Marine Pollution Bulletin 89 (2014) 49–58

Contents lists available at ScienceDirect

Marine Pollution Bulletin

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

Interannual heavy element and nutrient concentration trends in the topsediments of Venice Lagoon (Italy)

http://dx.doi.org/10.1016/j.marpolbul.2014.10.0360025-326X/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +39 041 234 8522; fax: +39 041 234 8582.E-mail address: brown@unive.it (B. Pavoni).

Mauro Masiol a,b, Chiara Facca a, Flavia Visin a, Adriano Sfriso a, Bruno Pavoni a,⇑a Dipartimento di Scienze Ambientali, Informatica e Statistica, Università Ca’ Foscari Venezia, Dorsoduro 2137, 30123 Venezia, Italyb Division of Environmental Health and Risk Management, School of Geography, Earth and Environmental Sciences, University of Birmingham, Edgbaston,Birmingham B15 2TT, United Kingdom

a r t i c l e i n f o

Article history:Available online 30 October 2014

Keywords:Heavy metalsNutrientsFactor analysisLagoonsVenice

a b s t r a c t

The elemental composition of surficial sediments of Venice Lagoon (Italy) in 1987, 1993, 1998 and 2003were investigated. Zn and Cr concentrations resulted in higher than background levels, but only Cd andHg were higher than legal quality standards (Italian Decree 2010/260 and Water Framework Directive2000/60/EC). Contaminants with similar spatial distribution are sorted into three groups by means of cor-relation analysis: (i) As, Co, Cd, Cu, Fe, Pb, Zn; (ii) Ni, Cr; (iii) Hg. Interannual concentrations are comparedby applying a factor analysis to the matrix of differences between subsequent samplings. A generaldecrease of heavy metal levels is observed from 1987 to 1993, whereas particularly high concentrationsof Ni and Cr are recorded in 1998 as a consequence of intense clam fishing, subsequently mitigated bybetter prevention of illegal harvesting. Due to the major role played by anthropogenic sediment resus-pension, bathymetric variations are also considered.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Heavy metals, generally described as elements with a density>5 g/cm3, such as lead (Pb), cadmium (Cd), zinc (Zn), mercury(Hg), chromium (Cr), copper (Cu), and iron (Fe), are among themain environmental threats to human health, being responsiblefor harmful effects depending on the type of metal, exposure time,concentration and degree of bio-accumulation in the food chain.Moreover, high contamination levels can seriously affect ecosys-tem equilibrium, compromising metabolic activities and commu-nity diversity. Trace concentrations of such elements areessential for organism physiology, but the development of industryand mining activities since the mid 19th century has rapidlyincreased metal pollution (Järup, 2003) to levels which may haveadverse effects on biota and ecosystem. Therefore its reduction isa serious challenge, necessary in order to protect human healthand to favour environmental conservation and recovery. However,background concentrations (naturally occurring) depend on localgeological features and an universal threshold value cannot beestablished for all ecosystems (Ridgway et al., 2003). This rendersthe assessment of actual pollution levels and the determinationof appropriate management policies very difficult. Despite the

recent increasing efforts to regulate and manage the environmen-tal impact of industrial discharges, past contamination still repre-sents a significant ecological risk. Specifically, in coastal areas,dredging may be a serious hazard for aquatic ecosystems, as sev-eral studies demonstrate that dredged sediments may be responsi-ble for increased toxicity and adverse ecological effects (seereferences in Han et al., 2011 and in Onorati et al., 2013 and refer-ences therein).

This study analyses the changes in heavy metal, carbon andnutrient levels in the surficial sediments of a coastal transitionalecosystem that has been affected for long time by significant con-taminant inputs due to urban, agricultural and industrial dis-charges. Since the mid 20th century, the ecosystem of VeniceLagoon (North-western Adriatic Sea) has been exposed to severalsubstantial anthropogenic pressures causing sediment pollution(Bellucci et al., 2000, 2002; Bernardello et al., 2006 and referencestherein; Guerzoni et al., 2007; Secco et al., 2005; Zonta et al., 2007),water contamination (Micheletti et al., 2011), eutrophication(Sfriso et al., 2003), overexploitation of biological resources(Pranovi et al., 2004), degradation of biota due to bioaccumulationof pollutants through the food chain (Raccanelli et al., 2004;Turetta et al., 2005; Sfriso et al., 2008), sediment erosion (Sfrisoet al., 2005; Sarretta et al., 2010; Rapaglia et al., 2011) and conse-quent salt marsh losses (Molinaroli et al., 2009). On May 2003, theItalian National Authorities formally started the construction of asystem to regulate tidal floods at the three inlets of the Lagoon

Fig. 1. (a) Map of Venice Lagoon, (b) Detail of the study area; (c) mud content (% pelite <63 lm); (d) Cadmium (lg cm�3); (e) Mercury (lg cm�3); and (f) Zinc (lg cm�3).NCB = Northern-Central Basin; SCB = Southern-Central Basin; CDP = Canale dei Petroli; CVE = Canale Vittorio Emanuele II; IDT = Isola delle Tresse.

50 M. Masiol et al. / Marine Pollution Bulletin 89 (2014) 49–58

Table 1Descriptive statistics of analysed parameters, elements and compounds in sediments sampled in 2003. Contamination factors were calculated with respect to pre-industrialbackground values in Pavoni et al. (1987).

Mean Std. Dev. Median Min. Max. Contamination factor Mean values in mg/kg to becompared with law qualitystandards (see Section 2.6)

Sedimentological parametersPelite % 62 31 65 6 98Density g cm�3 1.75 0.18 1.73 1.50 2.07

Major elementsAluminium mg cm�3 8.47 12.4 4.50 1.48 56.8Iron mg cm�3 15.3 2.52 14.9 10.5 22.9 0.72Manganese lg cm�3 288 53.4 276 192 459

Minor and trace elementsChromium lg cm�3 50.5 40.2 39.3 19.7 230 2.24 44.8Cobalt lg cm�3 7.59 1.92 7.47 3.51 11.4 0.48Nickel lg cm�3 17.7 14.7 13.2 6.38 82.6 0.79 15.8Copper lg cm�3 15.7 7.16 14.9 2.67 31.1 0.79Zinc lg cm�3 91.3 45.3 75.7 19.1 189 1.28Arsenic lg cm�3 9.43 2.01 9.00 6.83 15.1 0.89 8.89Cadmium lg cm�3 0.96 0.37 0.86 0.32 1.76 0.90 0.89Mercury lg cm�3 0.78 0.71 0.64 0.09 3.82 7.63 0.76Lead lg cm�3 11.8 9.48 11.7 0.80 30.0 0.50 12.5

Carbon, nitrogen and phosphorusTotal carbon TC mg cm�3 79.2 22.5 74.4 45.6 124Inorganic carbon IC mg cm�3 73.6 23.4 65.8 40.0 123Organic carbon OC mg cm�3 5.52 2.81 5.08 0.93 12.7Total nitrogen TN mg cm�3 0.69 0.29 0.75 0.13 1.15Total phosphorus TP lg cm�3 393 85.2 389 295 665Inorganic phosphorus IP lg cm�3 341 89.4 325 216 660Organic phosphorus OP lg cm�3 52.2 30.8 51.4 3.02 114

All values are given as dry weight (dw).

M. Masiol et al. / Marine Pollution Bulletin 89 (2014) 49–58 51

by means of mobile barriers, called MOSE. The system is supposedto be in operation in 2016, after the realization of 78 floodgates20 m wide, 18.5–29.6 m high and 3.6–5 m thick, depending onthe depth of the inlet (http://www.salve.it/uk). Beyond the realiza-tion of the barriers, several supporting interventions were made toadapt the inlet beds.

In the present paper, particular attention is paid to describe thecontaminant concentrations just at the beginning of the MOSE pro-ject (June 2003) to have a benchmark of lagoon conditions before aso important anthropogenic intervention that is expected to deeplychange lagoon’s dynamics. Moreover, multivariate statisticalapproaches were applied to simplify and summarise the informa-tion contained in a dataset of 22 variables (sediment texture anddensity: 12 elements including heavy metals; inorganic andorganic carbon; nitrogen; inorganic and organic phosphorus) and100 observations and find common patterns in pollutant inter-annual variations, focusing on bio-physical processes affectingcontaminant concentrations and distribution during 4 separatephases:

� June 1987: eutrophication due to massive Ulva proliferation wasthe main driver of biogeochemical cycles (Sfriso and Facca,2007); dystrophic–anoxic crises periodically occurred duringalgal degradation.� June 1993: macroalgal biomass almost disappeared (Sfriso and

Facca, 2007) mainly due to climate change and the synergiccombinations of many factors (Sfriso and Marcomini, 1996).� June 1998: uncontrolled clam harvesting peaked

(40,000 tons y�1, Pellizzato et al., 2011); hydraulic and mechan-ical dredging systems seriously affected the benthic habitat, byincreasing sediment resuspension (about one order of magni-tude, Sfriso et al., 2005), changing sediment texture and ampli-fying erosion processes (Molinaroli et al., 2009).� June 2003: at the beginning of the 2000s clam harvesting was

regulated by licensing �35 km2 for clam farming (Pellizzatoet al., 2011). The environmental impact significantly decreased

and seagrasses recolonized extensive areas of the lagoon (Sfrisoand Facca, 2007).

Metal concentrations were interpreted in relation to sedimentgrain-size and bathymetric modifications in the same period inorder to assess the link between changes in contaminationwith erosion or deposition processes. The relationship betweencontaminants and sediments is a key to understanding the fate ofheavy metals in the ecosystem.

2. Materials and methods

2.1. Study area

The Lagoon of Venice (Fig. 1a) is one of the few wetlands stillpresent on the northwestern coast of the Adriatic Sea (45� 24 N,12� 20 E) among those originating from the Flandrian transgressionca. 6000 years ago (Gatto and Carbognin, 1981; Brambati et al.,2003). It has a surface area of �550 km2 and is connected to theAdriatic Sea through the inlets of Lido (�900 m wide, 14 m deep),Malamocco (�600 m wide and 14 m deep with a depression of�53 m, the deepest in the northern Adriatic Sea) and Chioggia(�500 m wide and 8 m depth). Its average depth is ca. 1 m: only5% is deeper than 5 m, and 75% is shallower than 2 m (Molinaroliet al., 2009). Water and sediments are exchanged with the AdriaticSea following micro-tidal cycles which are semidiurnal, with anannual mean tidal range of ±31 cm and spring tide differences of�80 to +120 cm with respect to the mean tidal level. About 60%of the total volume is exchanged with the sea during each tidalcycle (12 h).

2.2. Sampling design

The 25 selected sampling sites (Fig. 1b) were located in shallowwaters in the central lagoon (about 1 m deep), distributed over

52 M. Masiol et al. / Marine Pollution Bulletin 89 (2014) 49–58

�130 km2 in the area most subject to pollution from effluents dis-charged by the Porto Marghera industries and urban sewage fromthe mainland (Mestre) and islands (Venice, Murano, Burano).Sediment samples were collected in June 2003 in the same stationsand in accordance with the same procedures reported in previousstudies carried out in 1987, 1993 and 1998 (Bernardello et al.,2006; Secco et al., 2005). At each site the upper 5 cm of 3 coresobtained by a Plexiglas corer (i.d. 10 cm) were carefully homoge-nized, immediately frozen (�20 �C) and then freeze-dried.

2.3. Elemental analyses

Freeze-dried sediment samples were pulverized using a plane-tary ball mill (Fritsch Pulverisette, Germany) and then 200 mg,weighed on a microbalance, were Microwave-acid digested intetrafluormethaxil vessels with 5 mL of MilliQ� water, 1.5 mL of48.9% HF (for trace metal analysis, J.T. Baker, USA), and 3 mL ofaqua regia – 69% HNO3 (Fluka TraceSELECT, Germany) and 37.5%HCl (J.T. Baker, USA) 1:3 v/v – in an Ethos 1600 labstation(Milestone, Italy) using two different programs. Details are givenin Supplementary material Table 1. After digestion, samples werediluted to 100 mL using MilliQ� water and stored in Teflon bottles.

The elemental analyses were conducted by inductively coupledplasma optical emission spectrometer (ICP–OES, Optima 5300 DV,Perkin–Elmer, USA), equipped with an S10 automatic sampler, aperistaltic pump, a cyclonic spray chamber, a quartz torch and analumina sample injection system. The operating conditions forthe instrument are summarised in Supplementary material Table 1.Mercury was analysed using a flow injection system analyser(FIMS-100, Perkin–Elmer, USA) and arsenic with a graphite furnaceatomic absorption spectrometer (model 4100 ZL, Perkin–Elmer,USA). The elements and the protocols were the same as those ofBernardello et al. (2006).

Reagent blanks were prepared and analysed together with thesamples, following the same procedures, and the obtained valueswere routinely subtracted. Single ionic standards for ICP were usedfor calibrating the instrumental response and to test linearity.Details are given in Supplementary material Table 1. The qualityand accuracy of quantitative analyses were routinely assured byanalysing a certified reference material (PACS-2, marine sediment)obtained from the NRCC (Canada). The recovery of each analysedelement ranged from 76% to 137%. Analytical precision was moni-tored by means of repeated analyses of both standards and sam-ples at different concentrations, the relative standard deviationfor each element being <5%.

2.4. Environmental parameter analyses

Sandy (grain size >63 lm) and pelitic (<63 lm) sediment frac-tions were separated by wet sieving and weighed after drying at110 �C (Bale and Kenny, 2005). Sediment density (g cm�3 d.w.)was determined by measuring the volume of replicate samplesand weighing after drying.

Data on nutrient concentrations in surficial sediments were dis-cussed in Facca et al. (2014) and used here in combination with thecontaminant values in statistical analyses. In particular, variablelist include inorganic and organic phosphorus (IP and OP, respec-tively), total phosphorous (TP = IP + OP), total nitrogen (TN),inorganic and organic carbon (IC and OC, respectively) andtotal carbon (TC = IC + OC). Details on methods can be found inSupplementary material.

2.5. Data normalization

Since the concentrations of elements such as metals may differfrom site to site as a result of natural geochemical characteristics, it

can be difficult to identify the role of anthropogenic sources. Forthese reasons, various normalization procedures are proposed inthe literature (Roussiez et al., 2005). The choice of normalizer,although of great importance, remains subjective, depending onlocal geomorphological characteristics. In the present study, alu-minium, as described by other authors (e.g. Roussiez et al., 2005),was not considered appropriate as it has long been discharged inthe lagoon. After careful consideration, dry sediment density(g cm�3) was chosen as the best compromise, since it is signifi-cantly correlated with grain size (Spearman’s correlation coeffi-cient q = �0.80, p < 0.001) and most of the considered elements(p < 0.05). This choice has also other advantages: it allows concen-trations to be expressed on a volumetric basis and highlights theactual element loads and patterns in relation to sediment features.

2.6. Bathymetric data

Bathymetric data were obtained during two campaigns carriedout by the Venice Water Authority (MAV) in the 1990s and 2000s.The first campaign was carried out between 1992 and 1993 usingboth single beam echo sounders and stereo restitution of aerialphotography. The second campaign was performed between1999 and 2002 using a multibeam bathymetric acquisition systemfor the main channels (depth > 5 m), a single-beam echo-sounderfor shallow waters and minor channels (depth < 5 m), stadia rodswith GPS for the areas near salt marshes and mudflats, and stereoaerial photography for natural and artificial salt marshes. Depth(m) is expressed with reference to the mean sea level as measuredin 1942 in Genova. As described in Molinaroli et al. (2009), bathy-metric data were interpolated using the inverse distance weightedalgorithm (power: 2, n� of points: 5), with an output cell size res-olution of 10 � 10 m. The map of the bathymetric changes wascomputed using a map algebra algorithm in ArcGIS: [(2000sbathymetry)–(1990s bathymetry)]. It was then corrected by takingaccount of the subsidence occurring between 1990 and 2000(Brambati et al., 2003).

2.7. Contamination factor and chemical status

The contamination factor (Cf) can be used to describe the levelsof contamination from toxic substances, in cases where their pre-industrial concentrations in the same study basin are available(Hakanson, 1980). It is calculated as:

Cf ¼ Ci=Cbg

where Ci is the measured concentration of a substance in the sedi-ment and Cbg is the pre-industrial geochemical background of thatsubstance. For the considered study case, the pre-industrial heavymetal background levels in lg g�1 dw are: Cr (20), Co (15), Ni(20), Cu (20), Zn (70), As (10), Cd (1), Hg (0.1), Pb (25), Fe (20,000)(Pavoni et al., 1987).

The European Water Framework Directive 2000/60/EC (WFD)was incorporated in the Italian Decree 260 (November 8, 2010)that established the chemical quality standards in the sedimentsof coastal and transitional waters (0.3 mg kg�1 d.w. for Cadmiumand Mercury, 30 mg kg�1 d.w. for Nickel and Lead, 12 mg kg�1 d.w.for Arsenic and 50 mg kg�1 d.w. for Chromium) and indicated thebackground levels as reference concentrations of ‘‘high status’’.

2.8. Chemometric approach

Before any multivariate treatment, variables were tested fordistribution normality using the Shapiro–Wilks and Kolmogorov–Smirnov and Lilliefors (p < 0.05) tests. Because results indicated anon-normal distribution and variables exhibited different kindsof skewness, a series of data pre-treatments were used following

Table 2Spearman’s correlation coefficients. Only significant correlations (p < 0.05) arereported. Highly significant correlations (p < 0.001) in bold face.

Pelite As Cd Co Cr Cu Fe Pb

As 0.44Cd 0.68Co 0.51 0.87 0.85CrCu 0.78 0.77 0.76 0.84Fe 0.63 0.79 0.59 0.86 0.77Hg 0.61 0.43Ni 1.00Pb 0.66 0.76 0.73 0.82 0.92 0.83Zn 0.60 0.77 0.89 0.89 0.93 0.66 0.85

Table 3Results of factor analysis on data from 2003 survey. Factor loadings of less than |0.4|are not shown, whereas values >|0.6| are in bold. Var (%): percentage of varianceexplained by each factor. Cum. Var. (%): cumulative variance.

Factor 1 Factor 2 Factor 3 Factor 4

Al 0.80 0.50As 0.50 0.51Cd 0.83Co 0.51 0.47 0.65Cr 0.94Cu �0.57 0.75Fe 0.80HgMn 0.91Ni 0.95Pb �0.66 0.58Zn 0.94TN �0.85IP 0.92OP �0.79IC �0.41 0.79OC �0.46 0.41Var (%) 36.0 28.8 10.6 7.50Cum Var (%) 36.0 64.9 75.5 83.0

Table 4Results of factor analysis on differences among variables. Factor loadings of less than|0.4| are not shown, whereas values >|0.6| are in bold. Variable abbreviations inTable 1. Var (%): percentage of variance explained by each factor. Cum.Var. (%):cumulative variance.

Factor 1 Factor 2 Factor 3 Factor 4

Al �0.42 0.45As 0.78Cd 0.82Co 0.84Cr 0.93Cu 0.90Fe 0.91Hg 0.73Mn 0.80Ni 0.95Pb 0.74 0.41Zn 0.88TN 0.44 �0.53IP 0.64 �0.46OP 0.57IC �0.80OC 0.49 0.58Var (%) 43.6 13.6 8.23 5.92Cum Var (%) 43.6 57.2 65.4 71.4

M. Masiol et al. / Marine Pollution Bulletin 89 (2014) 49–58 53

Reimann et al. (2002). A Box–Cox transformation was applied to allvariables to approach a normal distribution; a subsequent stan-dardisation (mean zero and unit variance) was also applied toovercome differences in variation ranges. After this series of data

transformations, the statistical tests were re-applied and resultsindicated normal distributions at p < 0.05 for most of the variables.The transformed dataset was thus used for statistical processing.

In order to point out the relationships among the variables andto highlight similar patterns in the spatial distribution of contam-ination in 2003, some bi- and multi-variate analyses were con-ducted using Statistica v. 10 (StatSoft). All spatial interpolationswere performed using the kriging interpolation algorithm andmaps were generated using QGIS 2.4.

The interannual changes in sediment contamination were mademore evident by using a series of chemometric tools: for each sam-pling site, differences in concentrations between subsequent sam-pling campaigns were computed, i.e. 1993–1987, 1998–1993 and2003–1998. Moreover, an additional fictitious sample with nulldifferences for all the variables was added to the dataset to be usedas a ‘‘zero value’’, i.e. no changes in the contamination for all thevariables. This new dataset (total 76 cases) includes negative val-ues (decreasing concentrations in two successive samplings) andpositive ones (increasing concentrations). Tests for normality,Box Cox transformation and standardization were carried out asdescribed above, after adding a constant to make all values >0.As this data transformation substantially normalised the testresults for most of the variables, this transformed dataset was usedas input for a Varimax rotated factor analysis (FA) to decrease thedata dimensionality and detect the main hidden processes causingvariation in heavy metal sediment content.

Along with factor loadings, FA also provided a n �m factorscores matrix: n cases (differences in concentrations) andm factors (new variables). However, given the above-describedpre-treatment of the input dataset, the zero factor scores, i.e. thosewith null differences in the concentrations between two successivesamplings, can be estimated to be similar to those obtained for theadditional fictitious sample with null differences for all variables.Corrected factor scores (cFS) were then obtained by subtractingfrom each factor score the related ‘‘zero’’ score. All spatial interpo-lations in this study were performed using the kriging interpola-tion algorithm.

3. Results and discussion

3.1. Sediment characteristics and contaminant concentrations (2003data)

Table 1 summarises the analytical results (all expressed as dryweight, dw) of the surficial sediments sampled in 2003. Mapsshowing the spatial distribution of mud content and heavy metals,exceeding law quality standards and background values are shownin Fig. 1. The maps of the distributions of all elements are availableas Supplementary material (Fig. SI1).

The distribution of fine sediments indicated a rather variablegrain-size distribution (Table 1) which is comparable with thosederived from a previous survey on sediment texture (Molinaroliet al., 2009). In the study area, there are two different geochemicaland mineralogical fingerprints (Stefani, 2002): (i) typical silicatecomponents (higher in the area south of Venice and in the innerlagoon parts) and (ii) carbonates (higher in the area north of Veniceand in the external lagoon parts). Most of heavy elements (Cr, Co,Ni, Zn, As, Cd, Pb) had lower concentration in the area character-ized by carbonates and higher where silicates are dominant.However, this can also depend on the location of industrialsources: the huge release of contaminants had always occurredin the area south of Venice.

Considering both chemical quality standards (sensu WFD) andbackground levels (Pavoni et al., 1987), the lagoon sediment aver-age concentrations for As, Co, Cr, Cu, Fe, Mn, Ni, Pb were below the

Fig. 2. Maps of changes in factors representative of heavy metal concentrations. Pelite changes between (a) 1998 and 1993; (b) 2003 and 1998. First factor changes between(c) 1993 and 1987; (d) 1998 and 1993; (e) 2003 and 1998. Second factor changes between (f) 1993 and 1987; (g) 1998 and 1993; (h) 2003 and 1998. Third factor changesbetween (i) 1993 and 1987; (j) 1998 and 1993; (k) 2003 and 1998.

54 M. Masiol et al. / Marine Pollution Bulletin 89 (2014) 49–58

Fig. 3. Box and whisker plots of heavy metal mean concentrations. Red lines indicate law threshold values and green lines indicate background values as in Pavoni et al.(1987). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4. Bathymetric maps for 1990s (a), 2000s (b) and changes between 1990s and 2000s (c).

M. Masiol et al. / Marine Pollution Bulletin 89 (2014) 49–58 55

reference values indicating high chemical status (sensu WFD).However, Cd and Hg values were above the law quality standards(Table 1; Figs. 1d and e). Moderate contamination above back-ground levels was observed for Zn (Table 1; Fig. 1f), which is notincluded in the regulations. Cr concentrations resulted higher thanbackground levels, but they were below the quality standards.

No sites were contaminated with Fe and Co. In the case of Niand Cr moderate contamination was observed only in the sur-roundings of the canal connecting the industrial area to the sea(Canale dei Petroli). Except for one site close to the Lido Inlet,all were ranked from moderately to highly contaminated with

Hg. It has been demonstrated that Hg contamination in VeniceLagoon is due to past industrial discharges, with only limitedinputs in recent years (Bloom et al., 2004). However, resuspensionof sediments, mainly due to anthropogenic activities, represent acritical factor in Hg cycling and distribution (Bloom et al., 2004),as disruption of the lagoon bed favours the distribution of con-taminated sediments over a wider area, thus explaining whythe level of Hg contamination is generally high. Arsenic, Cu andPb contamination was limited to sites close to the industrial zoneand the airport, whereas Cd and Zn contamination was morewidespread.

Table 5Spearman rank order correlations between the bathymetric differences (2000s–1990s corrected for the land subsidence) and the changes in the sediment contamination for theanalysed parameters. Bold face correlations are significant at p < 0.01. Abbreviations in Table 1.

Al As Cd Co Cr Cu Fe Hg Mn

�0.218 0.199 0.318 0.118 �0.071 0.342 0.282 0.603 0.355

Ni Pb Zn Sand TN IP OP IC OC

�0.075 0.525 0.233 �0.368 �0.068 0.271 0.18 �0.003 �0.251

56 M. Masiol et al. / Marine Pollution Bulletin 89 (2014) 49–58

3.2. Correlation (2003 data)

On a spatial scale, the relationships between the concentrationsof heavy metals and nutrients clearly emerge when consideringboth Spearman’s correlation coefficients (Table 2) and the FactorAnalysis results (Table 3). On the first axis (explaining 36.0% ofvariance) loadings >0.40 were recorded for Al, As, Co, Fe, Mn, andinorganic phosphorus, whereas on the second axis organicphosphorus, total nitrogen, Pb and inorganic carbon were the mostsignificant. Based on the correlation coefficients (p < 0.001) similarpatterns were identified in the distribution of As, Cd, Co, Cu, Fe, Pband Zn (Table 2). Hg contamination was correlated with no otherparameter and Cr and Ni were strongly correlated only with oneanother (q = 1.00, p < 0.001). This strong correlation is likely dueto the common pollution source of Cr and Ni, which is related tothe use of stainless steel in port infrastructures and ship hulls.These contain Ni and Cr in definite proportions and release themto the sediments when destroyed by corrosion substantially inthe same ratios.

Combining these results with the information on the metalsexceeding chemical quality standards it is possible to addressfuture investigations only on Cd, as tracer for most of heavy metals,Hg, as the element raising the major concern, and on Cr, that wasnot above law values but it was twice as high as background levels(Table 1).

3.3. Comparing the four sampling campaigns: interannual changes insediment contamination

On the interannual scale, Factor Analysis was key to summaris-ing the variability of all considered parameters and findingcommon patterns. The first four factors explained 71.4% ofcumulative variance (Table 4). Heavy metals fluctuated in accor-dance with similar patterns on both seasonal and spatial scales(Tables 2 and 4):

– most of the considered heavy metals were correlated with eachother, being significant (loading>|0.6|) along the first axis(Table 4).

– Cr and Ni were closely correlated (second factor).– Hg had independent variability (fourth factor).

The role of nutrient enrichment appeared to be negligible in theconsidered system, but inorganic phosphorus (IP) concentrationsfollowed the same pattern as heavy metals in the first factor.Inorganic carbon (IC) concentrations, which are related tosediment mineralogical features, followed a negative trend alongthe third axis. The change in IC concentrations was significantlycorrelated (p < 0.05) with the variation in pelite caused by clamharvesting, which amplified resuspension and erosion processes(Sfriso et al., 2005; Molinaroli et al., 2009). This further confirmsthe huge impact of clam harvesting on sediment characteristics.

Focusing on the changes in heavy metals, the maps in Fig. 2 andbox plots in Fig. 3 show the variations occurred from 1987 to2003. As a result of the significant reduction in activity in the

Porto Marghera industrial district, with the shutdown of manyplants and stricter regulations, from 1987 to 1993 factors 1 and 4displayed a general reduction and factor 3 no variation (Fig. 2).The same period saw the disappearance of the macroalgal bloomsresponsible for anoxia with aquatic fauna mortality and release ofreduced sulphur compounds (Sfriso and Facca, 2007). Thisfavoured the maintenance of oxidised conditions for several years,which in turn determined the entrapment of contaminants in thesediments. This positive trend leading to reduced contaminationwas interrupted by the intense and uncontrolled clam harvestingthat peaked in the late 1990s. From 1993 to 1998 a generalincrease in heavy metal concentrations was observed (Figs. 2 and3), which may be directly related to sediment disruption andresuspension caused by the hydraulic and mechanical devices usedin clam collection. Moreover, in the same period, the shippingchannels were enlarged as part of the renovation of cruise termi-nals and dredged materials and industrial tailings were used tobuild a framed artificial island in front of the industrial zone. Thishuge sediment handling operation worsened ecosystem condi-tions, reducing primary producers (Sfriso and Facca, 2007) andinterrupting the burial process of heavy metals, which remainedin the top layer or were redistributed over a wider area.

Finally, from 1998 to 2003, as a consequence of the decline ofclam stocks and the restrictions imposed on harvesting, the aver-age sediment concentrations of most heavy metals fell (Figs. 2and 3). However, another major cause of this reduction was thehuge sediment resuspension of the 90s. Most of the sedimentsneed hours before deposition, since, once re-suspended, the finestparticles (size 20–30 lm) require tidal and wind-driven currentsbelow 1 mm s�1 to settle (Tambroni and Seminara, 2005). Thesettling may occur far from the areas where the particles wereresuspended, and sediments may be lost to the sea on the ebb tide:Tambroni and Seminara (2005) estimated yearly net loss of finesand of up to 105 m3, or 106 m3 if the finest particles are included.Hence, it is reasonable to speculate that a significant amount ofcontaminants was also lost to the sea.

Only the fourth factor, Hg, showed no reduction, the contamina-tion level remaining very high (>6) (Hakanson, 1980) and abovethe law threshold (Fig. 3).

3.4. Relationship between contamination changes and bathymetricvariations

Fig. 4 shows bathymetric maps of the central lagoon in the1990s and 2000s and the changes between the two periods cor-rected for land subsidence. The results show that most of the cen-tral lagoon underwent moderate erosion (0.1–0.5 m), as alreadyreported by Sfriso et al. (2005), neutral areas (not affected bybathymetric changes) being located at the edges. In contrast, asexpected, the main canals and the area just north of the Lido inletaround the island of Sant’Erasmo experienced deposition of finesediments. The 1993–2002 bathymetric changes were processedin order to explore the relationships between contaminant changesand erosion/deposition trends. However, the Spearman’s coeffi-cients (Table 5) show that only Hg and Pb were significantly

M. Masiol et al. / Marine Pollution Bulletin 89 (2014) 49–58 57

(p < 0.01) and positively correlated with the bathymetric changes.One reason for the lack of a clear relationship between erosion/deposition processes and changes in the contamination levels formost of the considered elements is probably the re-distributionof contaminated sediments within the basin caused by clamfishing.

4. Conclusions

The contamination of Venice Lagoon was assessed by analyzingheavy metals, carbon and nutrient concentrations in surficialsediments collected at 25 sites in 2003. Most heavy metals(Fe, Mn, Co, Cu, Zn, As, Cd, Pb), organic carbon and total nitrogenshowed similar spatial distributions, having higher values in theinner part of the lagoon, probably due to both discharges fromthe industrial zone and drainage from urban areas. Mercury hasits own distinctive spatial distribution, with high concentrationsin the northern lagoon. Despite a slight decrease of Hg concentra-tion (not statistically significant one-way ANOVA p > 0.5), itremains the main factor responsible for the considerable degreeof contamination of the study area. Some attention must be paidalso to Cd, whose concentrations, despite declining, remain abovethe law threshold.

Even though the industrial activities related to these metalshave been discontinued (the chlor-alkaly plant (Hg) and the zincproduction from sphalerites were closed down and the electro-plating industry strongly reduced), contaminated sediments arere-distributed throughout the lagoon.

Factor Analysis returned a common pattern in the interannualand spatial variability of heavy metals, highlighting the followingmain issues: (i) the impact of sediment dredging on all elements;(ii) exceptions in the behaviour of Cr and Ni; (iii) high Hgcontamination.

In the late 1990s the effects of sediment resuspension caused byclam harvesting favoured the increase and re-distribution of con-taminants in the lagoon, but, in the following years, that processprobably determined a ‘‘dilution’’ of concentrations across a widerarea and loss to the sea. Combined with the reduced direct inputinto the lagoon, overall contamination displayed a decreasingtrend from the late 1980s till the early 2000s. A study on a 16-yearperiod supported by an extended chemometric approach enablessome target elements to be identified, that still remain a risk forhuman and ecosystem health, despite the interventions to mitigatethem. These are Cd, Hg and, to a lesser extent, Cr. This study con-tributes to the determination of baseline conditions in VeniceLagoon in view of the changes that are expected to result fromthe construction of the MOSE dams. Data on the occurring changesare being collected and will be published after the MOSE’s has goneinto operation. Following the described trends, in the future a fur-ther reduction due to the constant decrease of industrial activitiescan be expected, but no forecast on the effects of MOSE’s gates andinfrastructures can be attempted so far.

Acknowledgments

The authors are grateful to Consorzio Venezia Nuova(Dr. B. Bertani and Arch. G. Biotto) for providing bathymetric dataand to Dr. M. Marchiori (Regional Environmental ProtectionAgency, ARPAV) for mercury analyses. The authors are grateful toMr. George Metcalf for the English editing.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.marpolbul.2014.10.036.

References

Bale, A.J., Kenny, A.J., 2005. Sediment analysis and seabed characterization. In:Eleftheriou, A., McIntyre, A. (Eds.), Methods for the Study of Marine Benthos.Blackwell Science Ltd., Oxford, pp. 43–81.

Bellucci, L.G., Frignani, M., Raccanelli, S., Carraro, C., 2000. Polychlorinated dibenzo-p-dioxins and dibenzofurans in surficial sediments of the Venice Lagoon (Italy).Mar. Pollut. Bull. 40, 65–76.

Bellucci, L.G., Frignani, M., Paolucci, D., Ravanelli, M., 2002. Distribution of heavymetals in sediments of the Venice Lagoon: the role of the industrial area. Sci.Total. Environ. 295, 35–49.

Bernardello, M., Secco, T., Pellizzato, F., Chinellato, M., Sfriso, A., Pavoni, B., 2006. Thechanging state of contamination in the Lagoon of Venice. Part 2: Heavy metals.Chemosphere 64, 1334–1345.

Bloom, N.S., Moretto, L.M., Scopece, P., Ugo, P., 2004. Seasonal cycling of mercuryand monomethyl mercury in the Venice Lagoon (Italy). Mar. Chem. 91, 85–99.

Brambati, A., Carbognin, L., Quaia, T., Teatini, P., Tosi, L., 2003. The Lagoon of Venice:geological setting, evolution and land subsidence. Episodes 26, 264–268.

Facca, C., Ceoldo, S., Pellegrino, N., Sfriso, A., 2014. Natural recovery and policyinterventions of coastal wetlands: Venice Lagoon (Northern Adriatic Sea, Italy)as a case study. ScientificWorldJournal, <http://dx.doi.org/10.1155/2014/968618>.

Gatto, P., Carbognin, L., 1981. The Lagoon of Venice: natural environmental trendand man-induced modification. Hydrol. Sci. Bull. 26, 379–391.

Guerzoni, S., Rossini, P., Sarretta, A., Raccanelli, S., Ferrari, G., Molinaroli, E., 2007.POPs in the Lagoon of Venice: budgets and pathways. Chemosphere 67, 1776–1785.

Hakanson, L., 1980. An ecological risk index for aquatic pollution control. Asedimentological approach. Water. Res. 14, 975–1001.

Han, S., Gieskes, J., Obraztsova, A., Deheyn, D., Tebo, B.M., 2011. Relocation effects ofdredged marine sediments on mercury geochemistry: Venice Lagoon, Italy.Estuar. Coast. Shelf. Sci. 93, 7–13.

Järup, L., 2003. Hazards of heavy metal contamination. Brit. Med. Bull. 68, 167–182.Micheletti, C., Gottardo, S., Critto, A., Chiarato, S., Marcomini, A., 2011.

Environmental quality of transitional waters: the lagoon of Venice case study.Environ. Int. 37, 31–41.

Molinaroli, E., Guerzoni, S., Sarretta, A., Masiol, M., Pistolato, M., 2009. Thirty-yearchanges (1970–2000) in bathymetry and sediment texture recorded in thelagoon of Venice sub-basins, Italy. Mar. Geol. 258, 115–125.

Onorati, F., Mugnai, C., Pulcini, M., Gabellini, M., 2013. A framework for theintegrated assessment and management of dredged materials in Italy: a casestudy based on the application of Local Sediment Quality Guidelines. J. Soil.Sediment. 13, 474–487.

Pavoni, B., Donazzolo, R., Marcomini, A., Degobbis, D., Orio, A.A., 1987. Historicaldevelopment of the Venice Lagoon contamination as recorded in radiodatedsediment cores. Mar. Pollut. Bull. 18, 18–24.

Pellizzato, M., Galvan, T., Lazzarini, R., Penzo, P., 2011. Recruitment of Tapesphilippinarum in the Venice Lagoon (Italy) during 2002–2007. Aquacult. Int. 19,541–554.

Pranovi, F., Da Ponte, F., Raicevich, S., Giovanardi, O., 2004. A multidisciplinary studyof the immediate effects of mechanical clam harvesting in the Venice Lagoon.ICES J. Mar. Sci. 61, 43–52.

Raccanelli, S., Pastres, R., Favotto, M., Vio, P., 2004. Correlation between POPs insediment and edible bivalve in the Lagoon of Venice and estimation of the dailyintake. Organohalogen Comp. 66, 1823–1828.

Rapaglia, J., Zaggia, L., Ricklefs, K., Gelinas, M., Bokuniewicz, H., 2011. Characteristicsof ships’ depression waves and associated sediment resuspension in VeniceLagoon, Italy. J. Mar. Syst. 85, 45–56.

Reimann, C., Filzmoser, P., Garrett, R.G., 2002. Factor analysis applied to regionalgeochemical data: problems and possibilities. Appl. Geochem. 17, 185–206.

Ridgway, J., Breward, N., Langston, W.J., Lister, R., Rees, J.G., Rowlatt, S.M., 2003.Distinguishing between natural and anthropogenic sources of metals enteringthe Irish Sea. Appl. Geochem. 18, 283–309.

Roussiez, V., Ludwing, W., Probst, J.L., Monaco, A., 2005. Background levels of heavymetals in surficial sediments of the Gulf of Lions (NW Mediterranean): anapproach based on 133Cs normalization and lead isotope measurements.Environ. Pollut. 138, 167–177.

Sarretta, A., Pillon, S., Molinaroli, E., Guerzoni, S., Fontolan, G., 2010. Sedimentbudget in the Lagoon of Venice, Italy. Cont. Shelf. Res. 30, 934–949.

Secco, T., Pellizzato, F., Sfriso, A., Pavoni, B., 2005. The changing state ofcontamination in the Lagoon of Venice. Part 1: organic pollutants.Chemosphere 58, 279–290.

Sfriso, A., Facca, C., 2007. Distribution and production of macrophytes andphytoplankton in the lagoon of Venice: comparison of actual and pastsituation. Hydrobiologia 577, 71–85.

Sfriso, A., Marcomini, A., 1996. Decline of Ulva growth in the lagoon of Venice.Bioresource. Technol. 58, 299–307.

Sfriso, A., Facca, C., Ghetti, P.F., 2003. Temporal and spatial changes of macroalgaeand phytoplankton in a Mediterranean coastal area: the Venice lagoon as a casestudy. Mar. Environ. Res. 56, 617–636.

Sfriso, A., Facca, C., Marcomini, A., 2005. Sedimentation rates and erosion processesin the lagoon of Venice. Environ. Int. 31, 983–992.

Sfriso, A., Argese, E., Bettiol, C., Facca, C., 2008. Tapes philippinarum seed exposure tometals in polluted areas of the Venice lagoon. Estuar. Coast. Shelf. Sci. 79,581–590.

58 M. Masiol et al. / Marine Pollution Bulletin 89 (2014) 49–58

Stefani, C., 2002. Variation in terrigenous supplies in the Upper Pliocene to recentdeposits of the Venice area. Sediment. Geol. 153, 43–55.

Tambroni, N., Seminara, G., 2005. Long term net exchange of sands through Veniceinlets. In: Campostrini, P. (Ed.), Scientific Research and Safeguarding of Venice,vol. III. CORILA, Venezia, pp. 459–468.

Turetta, C., Capodaglio, G., Cairns, W., Rabar, S., Cescon, P., 2005. Benthic fluxes oftrace metals in the lagoon of Venice. Microchem. J. 79, 149–158.

Zonta, R., Botter, M., Cassin, D., Pini, R., Scattoli, M., Zaggia, L., 2007. Sedimentchemical contamination of a shallow water area close to the industrial zone ofPorto Marghera (Venice Lagoon, Italy). Mar. Pollut. Bull. 55, 529–542.