Water Research 36 (2002) 2491–2504

Assessment of river sediment quality by micropollutant analysis Marina Camussoa,*, Silvana Galassib, Davide Vignatia

a Water Research Institute, C.N.R., Via della Mornera 25, 20047 Brugherio, Milan, b University of Insubria, Via Valleggio 11, 22100 Como, Italy

Received 2 January 2001; accepted 19 October 2001

Abstract

Trace metals, PCB congeners and DDT homologues were determined in composite sediment samples collected from 10 representative sites along the river Po in two separate seasons. The aim was to identify the most anthropogenically impacted areas for future monitoring programmes and to aid development of Italian sediment quality criteria. The surface samples were collected during low flow conditions. Trace metal concentrations were assayed by electrothermal (Cd, Co, Cr, Cu, Ni, Pb), flame (Fe, Mn, Zn) or hydride generation (As) atomic absorption spectrometry after microwave assisted acid digestion. Hg was determined on solid samples by automated analyser. Organic microcontaminants were determined by gas-chromatography with 63Ni electron capture detector after Soxhlet extraction. Concentrations of trace metals, total PCB and DDT homologues showed two distinct peaks at the sites immediately downstream of Turin and Milan, respectively, and in each case decreased progressively further downstream. Principal component analysis identified three major factors (from a multi-dimensional space of 35 variables) which explained 85–90% of the total observed variance. The first and second factors corresponded to anthropogenic inputs and geological factors on sediment quality; the third included seasonal processes of minor importance. Sediment quality assessment identified Cd, Cu, Hg, Pb, Zn and organic microcontaminants as posing the most serious threats to river sediment quality. A reference site within the Po basin provided useful background values. Moderate pollution by organochlorine compounds was ascribed both to local sources and to atmospheric deposition. r 2002 Elsevier Science Ltd. All rights reserved.

Keywords: Organochlorine compounds; Trace metals; Sediment quality; River Po

1. Introduction lakes. However, in view of the persistence of many micropollutants and their potential for bioaccumula- Sediments are a recognised sink and reservoir for a tion, sediments are now regarded as an important source variety of environmentalcontaminants, includingboth of many micropollutants that seriously threaten natural naturally occurring substances subject to anthropogenic ecosystems [1]. Developing reliable methods for estimat- influences (e.g. nutrients and trace metals) and xenobio- ing the risks due to these substances in aquatic tic compounds (PCB, PAH, etc.). In the past attention environments has therefore become a priority. To concentrated on the role of sediments as a reserve complicate matters, many new chemicals have been compartment in the biogeochemicalcycleof phos- introduced into the biosphere so recently that their phorus, and as a source of phosphorus in eutrophic potential bioavailability and toxic effects are still unknown. Risk assessment must therefore be based on *Corresponding author. Tel.: +39-039-2004303; fax: +39- a few classes of inorganic and organic chemicals that 039-2004692. have been widely studied in both abiotic and biotic E-mail address: [email protected] (M. Camusso). matrices.

0043-1354/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0043-1354(01)00485-7 2492 M. Camusso et al. / Water Research 36 (2002) 2491–2504

Trace metals and organochlorine compounds are (September-October) 1996 and winter (February- among the most frequently monitored micropollutants, March) 1997. Samples from each site at each date were and reliable techniques have been established for their pooled to provide composite samples for analysis. extraction and quantification. The data available on Medium flow conditions prevailed on both sampling sediments have made it possible to propose sediment occasions, but discharge, measured at the Pontelagos- quality criteria and guidelines for these contaminants, curo (PLS) monitoring station, was more stable in which in some cases have been endorsed by environ- winter and two major high flow events (maximum mental protection agencies [2,3]. However, the reliability discharges at PLS=5570 and 6870 m3 s1) occurred of the numericalcriteria has not been establishedand between the two campaigns (Fig. 1). extensive field validation is necessary before these At each site samples were taken with a Ponar grab criteria can be routinely used for large-scale monitoring sampler along longitudinal transects of varying length [4]. Regrettably, natural background levels of trace (from 1–2 km in the upper stretches of the river to 20– metals and baseline pollution levels of xenobiotic 40 km in the lower reaches). The samples from each compounds are still unknown for many important transect were thoroughly mixed in situ under dim light freshwater ecosystems, and this often hinders the conditions and transported on ice to the laboratory [9]. establishment of reliable and widely applicable sediment After freeze-drying, whole sediment sub-samples were quality criteria. sieved through a standard sieve (ASTM no 230, Fritsch) All these problems apply to the river Po, the largest and the fraction less than 63 mm was employed for trace Italian river (and tenth largest European river in terms metal and organic micropollutant analysis [10]. Site 1 of length and basin area). Although studies on sediment (Monte Torino), only 75 km from the source and contamination in some stretches of the Po have been upstream of all major anthropogenic influences, served performed [5] and references therein [6–8]; systematic as reference site. All the other sites (2–10 in Fig. 1) were investigations have not been conducted along the entire downstream of the confluences of the Po’s main course. The present study, which is part of a compre- tributaries: Dora Riparia, Dora Baltea, Sesia, , hensive research project being carried out by the Italian Ticino, Lambro, Adda, Oglio and Panaro. In what Water Research Institute (IRSA-CNR), is the first follows, each sampling site is referred to mainly by the attempt to fill this gap. numbers in Fig. 1 but also by the name of the tributary. Spatialand seasonalvariations of trace metals,PCB As shown in Fig. 1, the study focused on the Alpine congeners and DDT homologues were investigated tributaries (left bank of the Po) rather than Apennine almost simultaneously (all sites were visited within 2 (right bank) tributaries since the former normally have weeks) at 10 representative locations evenly distributed much higher discharges, and much greater contaminant along the course of the river. Composite sediment loads. However, two Apennine tributaries were included samples were taken from each site in order to identify in the study: the Tanaro (site 5) which is the main the most anthropogenically impacted areas and also tributary in the upper Po and the Panaro (site 10) which identify priority compounds for future monitoring. is the last tributary of the Po. Downstream of the Overall sediment quality was also evaluated with respect confluence with the Panaro, the Po becomes estuarine to recently calculated numerical criteria. We are fully and its bed is above the level of the surrounding terrain. aware of the limitations of this type of evaluation which Site 10 is therefore in the closing section of the river. was performed with a view to developing Italian sediment quality criteria (as proposed by Italian 2.2. Trace metal analysis Legislative Decree 152, 11 May 1999). The findings of this study constitute the first set of homogeneous data Aliquots of approximately 0.1 g of sediment material for the contaminants investigated in the river Po. While (o63 mm) were weighed into acid-cleaned TFM vessels investigation of the time trends of sediment contamina- and digested with 5 mlof a mixture of nitric, hydro- tion in the river was beyond the scope of the study, it chloric and hydrofluoric acid (3:1:0.5) in a 1000W provides two time-separated but fully comparable microwave oven (Milestone 1200 MDR 1000/6/100/110) snapshots of sediment quality in the Po for some of operated at 35–50% of maximum power for 20 min. This the most toxic and persistent micro-contaminants. procedure proved suitable for the material from sites 6– 10 (lower Po) but did not completely digest sediments from sites 1–5 (upper Po). For the latter samples, 2. Materials and methods complete dissolution was achieved by adding 3.5 ml of a saturated solution of boric acid in hydrofluoric acid to 2.1. Sampling the standard acid mixture (EPA method 3052–95). After dissolution, the vessels were cooled to room temperature Sediment samples were collected at 10 sites along the and the solutions transferred to pre-conditioned poly- main course of the river (Fig. 1) in late summer thene vials, made up to 50 ml with deionised (Milli-Q) M. Camusso et al. / Water Research 36 (2002) 2491–2504 2493

Fig. 1. Sampling stations along the river Po: 1=Monte Torino (MT), 2=Dora Riparia (DR), 3=Dora Baltea (DB), 4=Sesia (SE), 5=Tanaro (TA), 6=Ticino (TI), 7=Lambro (LA), 8=Adda (AD), 9=Oglio (OG), 10=Panaro (PA). Flow regime measured downstream station 10 over the period September 1996FMarch 1997 is shown in the box.

water, and stored at 41C pending analysis. All digestions elements except Cd (15%). Closely similar RSDs (n ¼ 3) were performed in triplicate and procedural blanks were were found for the reference material, indicating that the routinely included to check for ambient or reagent adopted protocols were suitable for trace metals contamination. Trace metals were determined by elec- determination in the river Po sediments. trothermalatomic absorption spectrometry with deuter- ium background correction (Cd, Co, Cr, Cu, Ni and 2.3. Organic pollutant analysis Pb), air/acetylene flame absorption spectrometry (Fe, Mn and Zn), or hydride generation (As). Hg was Aliquots of 2 g of lyophilised sediment (o63 mm) were determined directly on solid samples on an automated extracted for 8 h in a Soxhlet with pesticide-grade n- Hg analyser (AMA254, FKV, Bergamo, Italy). hexane. Concentrated extracts (1–2 ml) were passed Recovery after digestion and mercury decomposition through a Florisil column (4 0.7 cm i.d.) with a layer was checked using a certified reference material(CRM (0.5 cm) of Cu powder on top, previously activated with 280: lake sediment) from the Community Bureau of HCl(18%) and washed with acetone and n-hexane. The Reference (Brussels). Percentage recoveries were always Florisil column was eluted with 25 ml of n-hexane and over 95% and no evidence of contamination was found the eluate was concentrated to 1 ml by rotary evapora- except, possibly, for Fe whose measured concentrations tion. were often about 110% of certified values. The Cleaned extracts (1 ml) were injected via an on-column reproducibility of the analytical procedure (n ¼ 5) was system into a gas-chromatograph (Carlo Erba GC 8000 checked on two river Po samples and the relative top) equipped with a 63Ni electron capture detector standard deviation (RSD) was less than 10% for all (Fison ECD 80). Operating conditions were: fused silica 2494 M. Camusso et al. / Water Research 36 (2002) 2491–2504 capillary column (50 m 0.25 mm i.d.), CP-Sil-8 CB of native approach is to calculate enrichment factors film thickness 1.9–2.0 mm; temperature programme from relative to levels at the reference site (site 1, MT). We 60 to 1801Cat201C min1, followed by 180 to 2701Cat used this method to verify quality assessment provided 1.51C min1. The carrier gas was helium at 1 ml min1 by the computer programme. Metal levels at a specific and the auxiliary gas was nitrogen at 30 ml min1; reference site were preferred to average shale values detector temperature was 3201C. because of the strong regionalvariabilityof metal Aroclor 1260 (Alltech), containing known quantities background levels [14]. of 18 PCB pure congeners (BCR, Brussels) (correspond- ing to 85% by weight of the totalPCBs present) was used as reference standard for PCB determination. A 3. Results and discussion reference standard mixture of DDT homologues was prepared from the pure compounds (Alltech). Sediment 3.1. Trace metals samples were extracted and analysed in duplicate and the overall precision of the analytical procedure Trace metalconcentrations in the fine sediment (extraction, purification and quantification of PCB fraction (o63 mm) during the two campaigns are congeners and DDT homologues) was about 15%. presented in Table 1. Co, Fe and Mn levels were Good laboratory practices were tested on a lyophilised essentially constant throughout the river Po, while for sludge sample certified for PCBs (BCR, Brussels). all the other elements there were moderate (2–3 fold) or marked (up to 6 fold) variations along the river course in 2.4. Statistical methods and sediment classification both sampling seasons. Levels of Cd, Cu, Hg and Zn procedures varied along the whole course of the Po; while As, Cr, Ni and Pb concentrations fluctuated markedly in the A Spearman correlation matrix was used to assess uppermost stretches of the river but were relatively interrelations between trace metals, PCBs, DDT homo- stable downstream of site 4. With a few exceptions logues, organic carbon (OC) and sediment texture (including the ‘‘stable elements’’ Co, Fe and Mn) characteristics. The best overall associations were maximum metal concentrations generally occurred at obtained by principalcomponent analysis(PCA) with sites 2, 7 or both. These are located downstream of the Varimax rotation of the component loading [11]. Since densely populated and highly industrialised areas con- PCA is highly sensitive to outlying observations [12], all taining the cities of Turin and Milan. However, the variables were log-transformed and the Shapiro–Wilk highest levels of Cd, Cu and Zn in the first campaign test used to check for normality. After transformation, occurred at site 9; while As was peculiar in that all variables except chromium and nickel in the first maximum values were at site 3 in the first campaign campaign were normally distributed (po0:01; Shapiro– and at site 1 in the second. Concentration minima Wilk test). However, PCA results were not significantly usually occurred at sites 5 or 6, except for As (site 2); altered by the inclusion or exclusion of these two highly while concentrations at site 1 were either comparable to skewed variables in the analysis and hence Cr and Ni or slightly higher than those at sites 5 and 6 (Tukey were included in the PCA for the first campaign. To honest standard difference test). However, site 1 avoid the risk of oversimplifying a complex environ- persistently had the lowest metal levels when differences mentalsituation, the PCA was first run on the trace in sediment texture between locations were taken into metals and organic microcontaminants separately and account (data not shown) and remains the reference site then for both sets combined. The two approaches for estimating metalenrichment factors in river Po yielded closely similar results and only the analysis on sediments. the whole data set is presented and discussed here. Marked seasonaldifferences in sediment concentra- An overall assessment of sediment quality was tions were exhibited by Cd and Hg at sites 3, 4 and 7; by performed using a specifically developed computer Cr, Ni, Pb and Zn at sites from 1 to 4; and by Cu and Zn programme [13]. Starting from experimentally measured at sites 9 and 10 (Table 1). At sites 1 and 3, some of these values, the programme calculates the corresponding differences were probably due to the marked changes in normalised concentration in a standard (reference) the percentage of the fine (o63 mm) sediment fraction sediment containing 25% of materialfiner than o2 mm between the two sampling periods [9]. However, and 10% organic matter (expressed as loss on ignition). seasonalchanges in the siltcontent of bed sediments The standardised concentrations are then compared to remained within 10% at all other stations except site 5 built-in sediment quality criteria (derived from Dutch (20% between the two campaigns), indicating that grain regulatory agencies) and a quality index ranging from 0 size distribution was only one of the factors controlling (no contamination) to 4 (heavy pollution) is calculated metalconcentrations in sediments of the river Po. for each contaminant. For trace metals, which have High winter Cd and Hg levels are probably linked to variable natural background concentrations, an alter- seasonalchanges in anthropogenic inputs, but for sites 3 M. Camusso et al. / Water Research 36 (2002) 2491–2504 2495

Table 1 Trace metalconcentrations ( mgg1 dry weight) in the fine fraction of the river Po sediments in two campaigns. The two-letter abbreviations identify the sampling sites: MT=Monte Torino, DR=Dora Riparia, DB=Dora Baltea, SE=Sesia, TA=Tanaro, TI=Ticino, LA=Lambro, AD=Adda, OG=Oglio, PA=Panaro. The numbers correspond to the site locations shown in Fig. 1

As Cu Ni Cr Co Cd Pb Fea Mn Zn Hg

(a) First campaign 1 (MT) 9.2 64 134 191 32.3 0.203 34.8 41.8 858 170 0.092 2 (DR) 6.6 97 283 396 39.2 0.958 63.8 42.4 1078 398 0.271 3 (DB) 19.5 72 118 194 33.7 0.338 32.2 46.4 1071 233 0.097 4 (SE) 17.9 66 97 163 36.9 0.284 27.0 51.5 1189 190 0.074 5 (TA) 16.3 61 91 154 32.0 0.198 22.5 45.7 1144 164 0.070 6 (TI) 14.6 50 101 167 28.3 0.261 18.5 41.2 1071 151 0.087 7 (LA) 11.6 84 104 177 25.5 0.465 26.3 38.1 1027 234 0.245 8 (AD) 11.3 65 106 191 25.6 0.465 26.9 35.0 892 244 0.214 9 (OG) 11.3 102 92 160 24.5 0.509 27.8 33.5 960 373 0.199 10 (PA) 9.8 76 85 148 23.3 0.345 25.5 32.4 985 296 0.168

(b) Second campaign 1 (MT) 14 56 130 205 28.6 0.272 18.2 42.6 771 214 0.090 2 (DR) 7.4 84 188 414 28.3 1.448 36.2 40.9 824 179 0.310 3 (DB) 11.3 82 152 299 30.2 0.849 27.7 41.3 765 263 0.170 4 (SE) 12.3 63 117 228 28.3 0.397 19.7 40.1 797 208 0.120 5 (TA) 12.8 44 83 176 23.1 0.215 14.3 39.9 834 187 0.092 6 (TI) 12.3 44 107 179 23.0 0.372 19.7 36.6 900 139 0.098 7 (LA) 11.4 96 109 169 20.8 0.794 27.2 34.6 713 244 0.327 8 (AD) 10.1 60 98 176 21.4 0.550 23.7 32.8 796 177 0.181 9 (OG) 9.4 69 90 149 21.3 0.522 23.8 33.4 882 212 0.178 10 (PA) 10.7 57 89 150 22.6 0.562 24.8 33.9 928 185 0.157

a 103.

and 4 also to remobilisation of polluted sediments from contamination in the river Po. However trace metal site 2. Late summer peaks of Cu and Zn at sites 9 and 10 concentrations reported for point sediment samples may originate from diffuse anthropogenic inputs from collected at sites 2 [18], 7 [7,6] and 10 [7,8] are very agriculture and stockbreeding [15], but may also be similar to values reported in Table 1, suggesting that linked to biological processes occurring in the lower metalcontamination in bed sediments of river Po has stretches of the Po. Pettine et al. [16] showed the been stable for at least the last 10 years. existence of a significant correlation between total dissolved phosphorous (taken as an indicator of 3.2. Organic micropollutants biological activity) and dissolved Cu and Zn concentra- tions in the Po at the closing station of Pontelagoscuro. Concentrations of organochlorine compounds in Po Further research is therefore required to verify the sediments are shown in Table 2. Only small differences influence of biological factors on metal concentrations in in PCB levels were observed between the campaigns at sediments of the lower Po, given that the role of each of the sites. Each pair of values was within the biological material as a scavenger for Cu and Zn has range of analytical variability (about 15%). DDT been clearly demonstrated for lacustrine environments homologues were determined only in the second [17]. The strong seasonalchanges of Cr, Ni, Pb and Zn campaign. pp’DDE was the main metabolite recovered levels in the upper Po are difficult to ascribe mainly to a in sediments, except for unexpectedly high values of single factor, and probably arise from the combined op’DDE at sites 2, 3 and 7. For totalPCBs and DDT effects of basin scale processes such as the overall homologues there were two distinct contamination hydrology of the river (particularly the regular occur- peaks: at sites 2 and 7, with progressively decreasing rence of autumn floods), the geology of the basin and concentrations downstream. Minimum concentrations fluctuations in anthropogenic inputs. Following these of total PCBs were found at site 1, and relatively low observations, it is clear that additional large-scale levels (about twice those at reference sites) were found in investigation are needed to better understand the factors the centralstretches of the river (sites 4, 5 and 6). Total controlling seasonal changes in sediment trace metal DDT followed a similar trend although between-site 2496 M. Camusso et al. / Water Research 36 (2002) 2491–2504

Table 2 PCBs and DDTs (ng g1 dry wt.) in the fine fraction of the river Po sediments in two campaigns. The numbers in the first column correspond to the site locations shown in Fig 1; site abbreviations are as in Table 1

1st campaign 2nd campaign

TotalPCB TotalPCB pp’DDT pp’DDE pp’DDD op’DDT op’DDE TotalDDT

1 (MT) 5.9 4.5 1.5 2.4 1.2 0.6 1.6 7.3 2 (DR) 74.7 69.2 2.8 4.9 1.8 0.5 7.7 17.7 3 (DB) 31.1 30.9 1.4 2.1 1.0 0.3 5.3 10.1 4 (SE) 13.0 12.7 0.7 0.8 0.4 0.1 1.9 3.9 5 (TA) 9.5 9.3 0.9 2.3 0.7 0.2 1.7 5.8 6 (TI) 10.2 10.3 0.6 1.4 0.7 n.d. 1.4 4.1 7 (LA) 76.1 70.7 3.1 5.2 1.7 0.3 8.4 18.7 8 (AD) 42.2 40.5 1.2 11.1 1.0 n.d. 2.2 15.5 9 (OG) 44.1 48.5 1.3 10.2 1.3 1.4 1.8 16.0 10 (PA) 25.1 23.1 2.3 6.7 0.8 0.1 1.5 11.4 n.d.=not determined.

differences were less pronounced than for PCBs. pp’ and Correlations were also found, on at least one sampling op’ DDE were the main components at almost all the occasion, between OC, Cu and PCB congeners; Fe and sampling sites. Co; Cr and Ni; and between PCB congeners and Cd, Cu, In a previous study [19] sediments collected in 1990 Hg, Pb and Zn. Correlations between DDT homologues from the terminalstretch of the river Po were analysed and other parameters were less significant and are not for PCB and DDT. Sediments at Pontelagoscuro (close shown in Table 3. to site 10) contained 75 ng g1 d.w. and 3 ng g1 d.w. of Based on the observed associations, iron oxides seem totalPCBs and pp’DDE, respectively.Other DDT to be the principalcarrier phases for some trace metals, homologues were under the detection limit (which was while OC is probably important (at least seasonally) for about the same as in the present study). TotalPCB at transporting Cu and PCBs. The strong relation of Cr site 10 in this study (average=24.1 ng g1 d.w.) was and Ni to the fine sediment fractions suggests that levels therefore lower than in 1990. However, the most of these two elements are strongly linked to the general polluted stations DR and LA (sites 2 and 7) had similar hydrology of the river which affects sediment grain size PCB levels to those found in the terminal reach of the Po distribution [10]. The high Cr and Ni concentrations at in 1990. As mentioned, these stations are downstream of site 2, and their marked spatialand seasonalvariations Turin and Milan and are therefore subject to strong at downstream sites in the upper Po, suggest that anthropogenic influences. Conversely, DDT pollution in autumn floods are able to transport material rich in Cr the finalstretches of the Po was higher than in 1990. and Ni from site 2 to downstream locations. Finally, the Furthermore, the presence of parent pp’ DDT and op’ correlation of Cu, Cd, Hg, Pb and Zn with the DDT, which were below detection limit in 1990, suggests xenobiotic PCBs, suggests that anthropogenic activity that the DDT input was recent and of industrialorigin. is an important source of these metals along the Po. It is possible that an industrial point source of DDT on a To identify factors possibly controlling these correla- tributary of Lake Maggiore [20,21] influenced the entire tions, best overall associations among the parameters River Po basin. were determined by principalcomponent analysis (PCA). The number of factors to extract from the 3.3. Relationships between investigated micropollutants multidimensional space of investigated variables was initially chosen following the Kaiser criterion; i.e. only A Spearman rank correlation matrix (Table 3) was those factors with an eigenvalue greater than one were constructed for trace metals, organic micropollutants, retained for further analysis. This equals to say that we sediment organic carbon (OC) and sediment grain size limited our attention to those factors which extracted at for the two sampling periods. In the first campaign the least the equivalent of the variance of one original o63 and o2 mm sediment fractions were both related to variable. This rationale resulted in a 3-factor model for OC, while Co, Cr and Ni only correlated with the the first campaign and in a 4-factor modelfor the second o2 mm fraction. In the second campaign the two grain- campaign. The scatterplots of factor loadings (after size parameters were strongly interrelated and correlated Varimax rotation) on components 1 and 2 (accounting with Co, Cr, Fe, Ni, and (for the finer fraction only) Mn. for some 80% of the totalvariance) are reported in M. Camusso et al. / Water Research 36 (2002) 2491–2504 2497

Table 3 Spearman correlation coefficients for the investigated variables in the two campaigns

First campaign 63m 2 mm2m OC * ++ OC As As Cu * Cu Ni * Ni Cr * * Cr Co * Co Cd * Cd Pb * * Pb Fe * Fe Mn ++ * Mn Zn * * Zn Hg * * * Hg PCBs ++ * * * *

Second campaign 63m 2m *2m OC OC As As Cu Cu Ni * * Ni Cr * * * Cr Co ++ * ++ * Co Cd * Cd Pb * * Pb Fe ++ * ++ * * Fe Mn ++ Mn Zn ++ Zn Hg * * * Hg PCBs * * * * DDTs ++ ++ ++

Statistically significant correlations: *=po0:01; ++=po0:05: 63 m=sediment fraction less than 63 mm in diameter; 2m=sediment fraction less than 2 mm in diameter; OC=organic carbon.

Fig. 2 and 3a for each parameter. The corresponding modelfor the second campaign does not affect the matrices of Varimax-rotated loadings, eigenvalues and relative positions of the investigated variables on the percent variances are presented in Tables 4 and 5, scatterplot of the first 2 factors (Fig. 3a and b). respectively. In both campaigns, the proposed model However, arsenic retains its direct relationship with explained approximately 90% of the total variance. grain size parameters and its inverse relationship with As to the second campaign, arsenic was the only Co, Cr, Fe and Ni irrespective of the modelof choice. variable for which the inclusion of the fourth factor in The environmentalinterpretation of the first 3 factors the PCA significantly increased the corresponding extracted by PCA would therefore be unaffected by the communality (from 0.16 to 0.87, data not shown). If choice of the model. For this reasons, we will limit arsenic was excluded from the analysis, the results ourselves to a 3-factor model also in the case of the obtained from the 4-factor model were practically the second campaign. This choice actually results in a more same as those obtained from a 3-factor modelincluding meaningfulinterpretation of experimentaldata (see: all variables (Fig. 3b). The most important difference sediment quality assessment) at the expenses of a very between the two alternate models is the change in the small part of the total variance extracted by PCA. signs of factor loadings on the second component of In both sampling periods, factor 1 included Cd, Cu, PCA; particularly important for Cr, Ni, Fe, Co and the Hg, OC, PCB congeners and DDT homologues (the grain size parameters (Table 5). With the obvious latter analysed only in the second campaign) with very exception of As, the choice of a three or four factor high loadings (>0.9) on all PCB congeners (Fig. 2 and 2498 M. Camusso et al. / Water Research 36 (2002) 2491–2504

First campaign 1 Cr Co 0.75 Ni Pb Cd 0.5 Fe Hg 0.25 Zn Cu 0 Mn Factor 2 Factor -0.25 Total PCB As -0.5 <63µm OC -0.75 <2µm

-1 -1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1 Factor 1 Fig. 2. Scatterplot of factor loadings (after Varimax rotation) for the investigated parameters in the first campaign.

Table 4 factor 2 in the first campaign and with factor 1 in the Results of Principal Component Analysis for the first campaign second campaign. Finally, factor 3 had high negative (factor loadings after Varimax rotation) loadings on Co, Fe, and Mn on at least one sampling Factor 1 Factor 2 Factor 3 occasion and positive loadings on OC and Zn in the second campaign. o63 mm 0.34 0.81 0.39 From the observed pattern of factor loadings, the first o2 mm 0.29 0.88 0.05 component of the PCA seems to summarise the overall OC 0.65 0.55 0.22 impact of anthropogenic activities on the quality of Po As 0.39 0.59 0.61 sediments. This is supported by the inclusion of OC in Cd 0.83 0.44 0.01 Co 0.16 0.66 0.70 factor 1 since pollution from organic matter is known to Cr 0.45 0.84 0.15 constitute a major threat to the overall quality of the Po Cu 0.80 0.14 0.11 ecosystem [8,22]. This interpretation suggests that Hg 0.86 0.14 0.42 anthropogenic inputs of Cd, Cu, Hg, Pb and Zn have Ni 0.34 0.89 0.13 the same origin as the xenobiotics. The exact sources of Pb 0.44 0.82 0.06 this contamination are difficult to identify since no Zn 0.81 0.17 0.20 major point sources of pollution are known in the Po Fe 0.40 0.28 0.83 basin. Diffuse inputs such as surface runoff from urban Mn 0.01 0.17 0.89 areas and disused industrialareas couldbe the main TotalPCB 0.96 0.04 0.09 source, and this supposition will be tested in subsequent Eigenvalue 17.97 5.11 2.97 studies. % Variance 61.9 17.6 10.2 The second component of the PCA includes metals that do not correlate with organic microcontaminants, and this probably reflects longitudinal variations in the geology of the river basin. Note, however, that Co and 3). Zinc had high loading on factor 1 only in the first Fe have negative loadings on factor 1 and remain well campaign. Factor 2 had high loadings on Co, Cr, Ni, separated from Cr, Ni and Pb which, by contrast, have and Fe in the second campaign; metals which are moderate positive loadings on factor 1 (Figs. 2 and 3). strongly associated with grain size parameters (Table 3). Anthropogenic activity may therefore have some influ- Arsenic also had a moderate negative loading (approx. ence on Cr, Ni and, seasonally, Pb levels in the upper 0.5) on factor 2, while Pb was mainly associated with Po, where Turin or metallurgical activities in the Dora M. Camusso et al. / Water Research 36 (2002) 2491–2504 2499

Table 5 Results of Principal Component Analysis for the second campaign (factor loadings after Varimax rotation)

A Factor 1 Factor 2 Factor 3 Factor 4 o63 mm 0.01 0.87 0.05 0.30 o2 mm 0.06 0.89 0.19 0.29 OC 0.57 0.10 0.60 0.10 As 0.01 0.40 0.02 0.84 Cd 0.83 0.50 0.03 0.04 Co 0.30 0.88 0.12 0.02 Cr 0.23 0.91 0.01 0.09 Cu 0.75 0.32 0.47 0.09 Hg 0.93 0.10 0.19 0.11 Ni 0.25 0.93 0.07 0.12 Pb 0.85 0.40 0.12 0.14 Zn 0.19 0.16 0.79 0.03 Fe 0.45 0.76 0.26 0.07 Mn 0.12 0.29 0.84 0.09 TotalPCB 0.94 0.02 0.10 0.11 pp’DDT 0.65 0.11 0.26 0.48 pp’DDE 0.62 0.47 0.15 0.47 pp’DDD 0.65 0.15 0.24 0.59 op’DDE 0.71 0.44 0.44 0.01

Eigenvalue 19.51 5.95 2.67 2.23 % Variance 57.4 17.5 7.9 6.6

B o63 mm 0.03 0.91 0.05 o2 mm 0.04 0.93 0.19 OC 0.59 0.08 0.59 As 0.14 0.57 0.06 Cd 0.81 0.49 0.02 Co 0.33 0.86 0.14 Cr 0.18 0.86 0.05 Cu 0.75 0.32 0.47 Hg 0.94 0.12 0.19 Ni 0.24 0.93 0.08 Pb 0.84 0.42 0.12 Zn 0.19 0.15 0.79 Fe 0.48 0.73 0.28 Mn 0.13 0.30 0.84 TotalPCB 0.94 0.00 0.10 pp’DDT 0.72 0.21 0.22 pp’DDE 0.70 0.36 0.21 pp’DDD 0.74 0.27 0.18 op’DDE 0.69 0.42 0.46

Eigenvalue 19.99 6.15 2.64 % Variance 58.8 18.1 7.8

A: four factor model; B: three factor model.

Riparia basin probably represent the ‘‘local’’ pollution identify the origin of these elements in the upper Po sources [15]. On the other hand, very high natural levels remain a priority. of Ni (up to 400 mgg1) and Cr (up to 300 mgg1) have The third component of PCA probably reflects the been measured in some small tributaries of the Dora presence of additionalprocesses which occur in the basin Riparia basin [18]. Since Cr and Ni concentrations in of the river Po on a seasonalbasis and/or on a local this part of the Po are much higher than any proposed scale. In the first campaign, factor 3 probably accounts sediment quality guideline, additional investigations to for local hydrological and geological factors which result 2500 M. Camusso et al. / Water Research 36 (2002) 2491–2504

Second campaign (Four factor model) 1 <63µm 0.75 <2µm pp'DDE 0.5 Mn As 0.25 OC Total PCB 0 Zn pp'DDT Hg Factor 2 Factor -0.25 Cu pp'DDD Pb -0.5 op'DDE Fe Cr Ni Cd -0.75 Co -1 -1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1 (a) Factor 1 Second campaign (Three factor model) 1 Co Ni Fe Cr 0.75 op'DDE Cd 0.5 pp'DDD Cu Pb 0.25 Zn pp'DDT Hg

0 Total PCB Factor 2 Factor OC -0.25 Mn As -0.5 pp'DDE

-0.75 <63µm <2µm -1 -1 -0.75 -0.5 -0.25 0 0.25 0.5 0.75 1 (b) Factor 1 Fig. 3. Scatterplot of factor loadings (after Varimax rotation) for the investigated parameters in the second campaign. (a) Four factor model; (b) three factor model.

in the As and Fe maxima at sites 3 and 4, respectively. In 3.4. Sediment quality assessment the second campaign, factor 3 highlights changes in the relative importance of organic matter originating from Comparison of the relative importance of factors 1 natural sources (allochthonous and autochthonous) and (man-made contamination on a basin scale) and 2 (basin from wastewater treatment plants [9]. These two sources geology and/or local pollution sources) at different sites are more differentiated in the winter period after the can help to identify areas where sediment pollution is occurrence of the autumn floods. Similarly, Zn contam- problematic (Fig. 4). At the same time, the scores on ination seems to originate from two sources as suggested factor 3 give an indication of those sites where seasonal by the seasonalpattern of spatialvariation of this and local processes have a greater role in controlling element in the two campaigns (Table 1). sediment quality (data not shown). M. Camusso et al. / Water Research 36 (2002) 2491–2504 2501

First campaign 2.5 2 2

1.5 1 1

0.5 4 8 0 6 3 Factor 2 Factor -0.5 5 9 10 7 -1

-1.5

-2

-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 Factor 1 Second campaign 2.5

2 2 1.5 3 1 1

0.5 4 6 0 8 Factor 2 Factor -0.5 7 5 10 -1 9 -1.5

-2

-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 Factor 1 Fig. 4. Scatterplot of scores for the ten sampling stations in the two campaigns. 1=Monte Torino (MT), 2=Dora Riparia (DR), 3=Dora Baltea (DB), 4=Sesia (SE), 5=Tanaro (TA), 6=Ticino (TI), 7=Lambro (LA), 8=Adda (AD), 9=Oglio (OG), 10=Panaro (PA).

On both sampling occasions, sites 2 and 7 had the stretches. Finally factor 3 showed positive scores at sites highest scores on factor 1 followed by sites 9 and 8 (the 1, 8, 9 and 10 in the first campaign and at sites 1, 3, 4, 5 latter in the second campaign only). All the other sites and 7 in the second campaign. had either low positive (Dora Baltea, Panaro and Adda, These findings identify sites 2 and 7 as those most the latter in the first campaign only) or negative (Ticino, heavily impacted by human activities, even though they Tanaro, Sesia and Monte Torino) scores on this factor. appear to have very different geologies (see above By contrast, factor 2 showed a more regular pattern, discussion on Ni and Cr). Furthermore, the two polluted with high scores in the uppermost stretches of the river sites have similar scores on factor 3 in the first campaign, and stable, moderately negative values in the lower but not in the second (data not shown). On a seasonal 2502 M. Camusso et al. / Water Research 36 (2002) 2491–2504 basis, the origin of sediment pollution may then differ 1000 m3 s1) where the river has greater capacity to between these two areas. By contrast, sites 1, 4, 5 and 6 buffer major pollutant inputs. Nonetheless the pollutant seem unaffected by anthropogenic activities (negative load entering the Po from the Lambro basin seems to be scores on factor 1) but, on factor 2, site 1 remains so large as to prevent complete recovery of sediment distinct from the others which form a homogeneous quality in the final part of the Po. Thus the ‘‘Lambro group especially in the first campaign (Fig. 4). This effect’’ [6] is also evident for organic microcontaminants, behaviour strengthens the hypothesis of geological Pb, Cd, Hg and, on one sampling occasion, Cu and Zn differences between the upper Po and the rest of the of the sediments. river basin. Trace metal, PCB and DDT levels at sites 4, 5 and 6 Sites 8, 9 and 10, downstream of the river Lambro were comparable to those measured at site 1, although inlet, were grouped by factor 1, consistent with a major for DDT homologues site 1 was not the least polluted. anthropogenic input being progressively diluted. This is Atmospheric deposition and trapping in high altitude an illustration of the well-known ‘‘Lambro effect’’ [6] on snow are probably the responsible for the relatively high Po waters, although the greater contamination at site 9 levels of DDT homologues (which are more volatile than than site 8 suggests that additional unidentified polluting PCBs) at this site. sources are present in the lower stretches of the river. The results produced by the sediment quality assess- However, site 10 had a very low score on factor 1 ment computer programme are shown in Table 6. The indicating that effective dilution of polluting inputs is programme did not identify any site as ‘‘class 4’’ (heavily generally achieved before the river enters the delta. polluted) indicating absence of gross sediment pollution Finally, for site 3, the score on factor 1 indicates no in the river Po. Among trace metals, Ni was class 3 major anthropogenic impact in either sampling periods, (medium–high pollution) for all sites except site 1. Cu while the variations on factor 2 and 3 are probably was class 2 (medium–low pollution) at sites 9 and 10 in related to the changes in grain size distribution and sedi- both sampling periods and at site 3 in the first campaign, ment type which occurred between the two campaigns. and sites 2, 4, 7, and 8 in the second campaign. Cr and The similar behaviour of sites 3 and 8 in the first Zn were enriched above background levels (class 1) at campaign is interesting. These two stretches of the river some stations with slight differences between the two seem to have similar geologies and anthropogenic sampling periods (Table 6). The programme did not influences, but the quality of their sediments can be indicate contamination by Cd, Hg, Pb or As; it did not significantly changed by resuspension and subsequent consider Co, Fe or Mn as there were no specifications deposition of material from polluted upstream sites (2 for these metals. The programme indicated that and 7). However, both inorganic and organic contam- sediment contamination by Ni and Cu is widespread ination declined sharply downstream of site 2, but not and relatively severe in the Po, while Cr and Zn from site 7 (Lambro). The difference may be due to contamination are important locally. Similar results differing flow regimes. At site 2 the Po has an average (data not shown) were obtained comparing the mea- discharge of 50–70 m3 s1 and may be unable to sured metalconcentrations with other numericalguide- effectively dilute a concentrated, though quantitatively lines for sediment quality [2,23]. limited, contaminant input. By contrast site 7 is on the The conclusions arising from the computer pro- large potamal section of the Po (average discharge about gramme assessment were in good generalagreement

Table 6 Assessment of sediment quality in the river Po in the second campaign (winter 1997) using the WaBoos computer programme [13]. PCBs are identified by the number of the congener; DDTs indicates the sum of pp-DDT, pp-DDD, pp-DDE, op-DDT and op-DDE. With few exceptions (see text) the same classification applies to the first campaign. See Table 1 for identification of sites and Fig. 1 for their location

SITE Class 0 Class 1 Class 2 Class 3

(1) MT As, Cd, Cr, Cu, Hg, Pb, Zn, 138, 153, 180 DDTs Ni (2) DR As, Cd, Hg, Pb, Zn Cr Cu, 101, 138, 153, 180, DDTs Ni (3) DB As, Cd, Cu, Hg, Pb, Zn Cr 101, 138, 153, 180, DDTs Ni (4) SE As, Cd, Hg, Pb, Zn Cr, DDTs Cu, 101, 138, 153, 180 Ni (5) TA As, Cd, Cu, Hg, Pb, Zn Cr 101, 138, 153, 180, DDTs Ni (6) TI As, Cd, Cr, Cu, Hg, Pb, Zn, 101 DDTs 138, 153, 180 Ni (7) LA As, Cd, Cr, Hg, Pb Zn Cu, 101, 138, 153, 180 Ni, DDTs (8) AD As, Cd, Cr, Cu, Hg, Pb, Zn Cu, 101, 138, 153, 180 Ni, DDTs (9) OG As, Cd, Cr, Hg, Pb, Zn Cu, 101, 138, 180 Ni, 153, DDTs (10) PA As, Cd, Cr, Hg, Pb, Zn Cu, 101, 138, 153, 180 Ni, DDTs M. Camusso et al. / Water Research 36 (2002) 2491–2504 2503 with the PCA (Tables 4–6). In fact the two approaches The high DDT levels in sediments at most sites had an are complementary since PCA identifies the general important negative effect on sediment quality. However patterns of sediment contamination, while sediment the qualitative aspects of this pollution are particularly quality guidelines are necessary to identify the proble- worrying since pp’DDT, banned for agricultural use matic contaminants at each site. However, evaluation of since the 1970s and at one time absent from Po metalcontamination with respect to the reference site 1 sediments, has now returned together with its op’ (Monte Torino) led to different conclusions. isomerFstrongly suggesting the new contamination is Comparison of metal levels at downstream sites with of industrialorigin. those at the reference site indicated that Cd and Hg enrichment was the most severe and most widespread in the Po sediments, followed by Cu and Zn; while Cr, Ni 4. Conclusions and Pb were significantly enriched only at sites 2 and 3. Conversely Fe, Mn and Co enrichment was negligible in (1) Sediment contamination along the river Po showed Po sediments, indicating that their concentrations in the distinct peaks at sites 2 and 7 downstream of Turin Po are close to natural background levels. Cr and Ni and Milan, respectively. may pose an environmental risk locally (at site 2) in the (2) The high contamination at site 2 has mainly a local Po basin; but we feelthat they do not pose the impact compared to contamination at site 7, since widespread risk suggested by application of the numer- sediment quality at sites 4–6 was comparable to that ical guidelines. The studies of Farini et al. [18] and at site 1 (reference site). By contrast, moderate Diserens [24] support the hypothesis of naturally high contamination persisted in the finalpart of the river background levels of Cr and Ni in the upper Po basin. (after site 7). Diserens [24] showed that more than 50% of river Cr (3) Although the major polluting inputs enter the Po was immobilised in the crystalline lattice of sediment from the areas around Turin and Milan, additional minerals and hence was not available to living organ- inputs of metals, PCBs and DDTs seem to come isms. It is noteworthy that the reference site comparisons from other tributaries or other diffuse sources in the are in better agreement with the order of impact of trace lower Po. Future studies should therefore assess metals on natural element cycles [25] which is as follows: areas not investigated in the present work. (4) Organic micropollutants, Cd, Hg, Pb and, to a Cd > Pb ¼ Hg > Cu > Zn > Cr > Ni > Fe ¼ Mn: lesser extent, Cu and Zn are the ‘‘priority con- With regard to organic micropollutants, only a limited taminants’’ that require continuous monitoring in number of PCB congeners (101, 138, 153 and 180) could the Po. However, widening the quality assessment be entered into the computer programme; while the to other pollutant classes and validation by concentrations of all the quantified DDT homologues ecotoxicological tests on benthic organisms is were automatically summed to yield a single quality desirable for environmental protection and manage- value (Table 6). PCBs were at the class 2 pollution level ment purposes. at all sites except site 1, indicating moderate but (5) Determination of the extent to which the high levels widespread contamination of Po sediments by these of Cr and Ni in upper Po sediments are natural compounds. Contamination by DDT homologues was remains a priority, since the measured concentra- low (class 1) at sites 1, 4 and 6, medium (class 2) at sites tions of these two elements are much higher than 2, 3 and 6, and medium-high (class 3) downstream of the sediment quality guidelines or criteria. river Lambro (sites 7, 8, 9 and 10). (6) The results of the present study suggest that the Risk assessment for PCB accumulation in fish was sedimentological approach is a useful one for evaluated in a previous paper [26] starting from identifying polluted tracts of a large river. We sediment pollution and assuming equilibrium between recommend the use of composite sediment samples the sediment and the water column. This model was to avoid sampling errors arising from the natural subsequently validated for aquatic trophic chains in the heterogeneity of the sediments. However, point terminalstretch of the Po [19]. Both theoretical samples may be useful for identifying hot spots in calculations and direct measurements demonstrated areas with degraded sediment quality. that, in the terminalstretch, some fish species are at risk of impaired reproduction due to PCB tissue accumulation. This risk may have decreased in recent years as we found some sediment recovery at site 10. Acknowledgements However, the sediments analysed in the present work were pooled samples while previous studies used The authors thank RaulVenema for providing a copy individual samples which would identify pollution hot- of the sediment quality assessment software package spots. WaBoos 05. 2504 M. Camusso et al. / Water Research 36 (2002) 2491–2504

References [14] Matschullat J, Ottenstein R, Reimann C. Geochemical backgroundFcan we calculate it? Environ Geol [1] Sodergren A. Trends in water pollution, consequences for 2000;39(9):990–1000. ecotoxicology. In: Zelikoff JT, editor. Ecotoxicology: [15] Autorita" di Bacino del Fiume Po, Bulletin of Po River responses, biomarkers and risk assessment, an OECD Basin Authority (Italian). 1997;19/20: 1–9. workshop. Fair Haven, NJ, USA: SOS Publications, 1997. [16] Pettine M, Camusso M, Martinotti W, Marchetti R, p. 15–23. Passino R, Queirazza G. Soluble and particulate metals in [2] Smith SL, MacDonald DD, Keenleyside KA, Ingersoll the Po river: factors affecting concentrations and parti- CG, Field LJ. A preliminary evaluation of sediment tioning. Sci TotalEnviron 1994;145:243–65. quality assessment values for freshwater ecosystems. [17] Sigg L. Metal transfer mechanisms in lakes; the role of J Great Lakes Res 1996;22(3):624–38. settling particles. In: Stumm W, editor. Chemical processes [3] Long ER, MacDonald DD. Recommended uses of in lakes. New York: Wiley, 1985. p. 283–310. empirically derived sediment quality guidelines for marine [18] Farini A, Vigano" P, Gigliotti C. Heavy metals in sediments and estuarine ecosystems. Hum EcolRisk Assess as indicators of anthropogenic impact. Zn, Cu, Ni, Cr 1998;4:1019–39. contents in sediments from Rivers Dora Riparia, di [4] Chapman PM, Wang F, Adams WJ, Green A. Appro- Lanzo and Po. Acqua Aria 1988;3:339–46 [in Italian]. priate applications of sediment quality values for metal [19] Galassi S, Guzzella L, Battegazzore M, Carrieri A. and metalloids. Environ Sci Technol 1999;33(22):3937–41. Biomagnification of PCBs, pp’DDE and HCB in the river [5] Galassi S, Provini A. Heavy metals and organic micro- Po ecosystem (Northern Italy). Ecotox Environ Safety pollutants in sediments and biota from River Po. Acqua 1994;29:174–86. Aria 1993;6:619–25 [in Italian]. [20] Ceschi M, DeRossa M, Jaggli M. Organic and inorganic [6] IRSA. Lambro-Po System: Transport of Pollutants and pollutants and radionuclides in the ichthyofauna of the Biological Effects. Quad Ist Ric Acque 1997;102:442pp. Lakes Ceresio and Verbano (Swiss basins). Trav Chim [in Italian]. Aliment Hyg 1996;87:189–211 [in Italian]. [7] Camusso M, Balestrini R, Martinotti W, Arpino M. [21] Bacchetta R, Binelli A, Galassi S, Provini A, Vailati G. Spatialvariations of trace metaland stableisotope content Contamination effects of DDT in Lake Maggiore on in autochthonous organisms and sediments in the Po river Dreissena polymorpha spawning. Acqua Aria 2000;8:105– (Italy). Aquat Ecosystem Health Manage 1999;2:39–53. 11 [in Italian]. [8] Camusso M, Vignati D, van de Guchte C. Ecotoxicologi- [22] Camusso M, Pagnotta R. Water quality state of the River calassessment in the rivers Rhine (the Netherlands) and Po Po basin. In Proceedings Symposium ‘‘Research perspec- (Italy). Third International Symposium on Sediment tives in water ecology’’, Milan, 20–21 June 1996. Quad Ist Quality Assessment, Amsterdam, August 16–19, 1998. Ric Acque 1997;103:152–87 [in Italian]. Aquat Ecosystem Health Manage 2000;3:335–45. [23] Guchte van de C, Mulder MAAJ. Lowest reported [9] Vigano" L. Assessment of the toxicity of River Po sediments NOEC’s for the effects of heavy metals, organochlorine with Ceriodaphnia dubia. Aquat Toxicol2000;47:191–202. pesticides and polycyclic aromatic hydrocarbons on [10] Westrich B. Hydromechanic aspects of contaminated Chironomus riparius, Daphnia magna and Photobacterium sediment transport in fluvialsystems. In: SlyPG, editor. phosporeum. Internaldocument, Institute for InlandWater Sediments and water interactions. New York: Springer, Management and Wastewater Treatment (RIZA), 1986. p. 63–8. Lelystad, The Netherlands, 1995, 15pp. [11] Digby PGN, Kempton RA. Multivariate analysis of [24] Diserens J. Natural, anthropogenic influences on chemical ecological communities. London: Chapman & Hall, 1986. composition of sediments and particulate matter from the p. 206. Po (Italy). Diplome# d’etudes! superieures! en sciences [12] Reinemann C, Filzmoser P. Normal and lognormal data naturelles de l’environnement, University of Geneva, distribution in geochemistry: death of a myth. Conse- 1997, pp. 79 [in French]. quences for the statisticaltreatment of geochemicaland [25] Nikiforova EM, Smirnova RS. Metaltechnophilityand environmentaldata. Environ Geol2000;39(9):1000–14. lead technogenic anomalies. Proceedings of the Interna- [13] Bakker T, Venema R. Waterbodem BOOS, version 5. tionalConference on Heavy Metalsin the Environment, Institute for Inland Water Management and Waste- Toronto, 1975, C94–96. water Treatment (RIZA), Department of Chemistry [26] Galassi S, Migliavacca M. Organochlorine residues in and Ecotoxicology, Lelystad, The Netherlands, 1997 River Po sediments: testing the equilibrium condition with [in Dutch]. fish. Ecotox Environ Safety 1986;12:120–6.