You are on page 1of 5

Small Ruminant Research 106S (2012) S49–S53

Contents lists available at SciVerse ScienceDirect

Small Ruminant Research
journal homepage:

Sheep milk as a potential indicator of environmental exposure to dioxin-like polychlorinated biphenyls (dl-PCBs)ଝ
˜ a , M. Perugini a,∗ , M. Esposito b , L. Baldi c , M. Amorena a E.G. Herrera Nunez
a b c

Dipartimento di Scienze degli Alimenti, University of Teramo, Via Crispi 212, 64100 Teramo, Italy IZS del Mezzogiorno, via Salute 2, 80055 Portici (Na), Italy ORSA Osservatorio Regionale Sicurezza Alimentare, via Salute 2, 80055 Portici (Na), Italy

a r t i c l e
Keywords: Dioxin-like PCBs Sheep milk GIS

i n f o

a b s t r a c t
Dioxin-like polychlorinated biphenyls are lipophilic persistent organic pollutants which accumulate in the fat tissue of animals. Moreover, in mammals, these contaminants are secreted into milk during lactation periods. The contamination level of milk is closely related to the animal dl-PCBs exposure through diet, inhalation and dermal contact. The study of the relationship between the environment in which the animals live and the milk contamination level is a key component in assessing the presence of pollutants in dairy products. In this paper, soil was selected as a reliable matrix of environmental pollution with dlPCBs as it reflects as well the possible contamination of vegetation and air in the areas where sheep graze and live. However, dl-PCBs analysis could be problematic due to the high cost. To overcome this problem, the approach of this study was focused on using GIS and geostatistics to integrate data from different research institutions, aiming to assess the relationship between milk and soil contaminated with dl-PCBs. Therefore, samples of soil and sheep milk collected in Campania (Italy) during 2008 monitoring plans were introduced in a GIS. The relationship between the two variables showed heteroscedasticity, and the global regression technique OLS indicated a R2 of 0.24. To improve the OLS results, the altitude was integrated in the GIS considering that this variable can influence PCBs contamination. The results showed that altitude is strongly related with soil contamination but was not strongly related with milk contamination. Thus, analyzing the relationship between milk and soil contamination dividing data in three different groups of altitude ranges, the R2 values were higher. Furthermore, this correlation changes with the altitude, becoming weaker with higher altitude. These results show the possibility to use sheep milk as indicator of PCBs contamination at least from sheep bred under 300 m of altitude. © 2012 Elsevier B.V. All rights reserved.

1. Introduction The dioxin-like PCBs (dl-PCBs) are a group of 12 polychlorinated biphenyls, showing chemical and toxicological

ଝ This paper is part of a supplement titled “SIPAOC Congress 2010” guest edited by Antonello Carta, Maria Teresa Manfredi, Giuseppe Moniello and Massimo Trabalza-Marinucci. ∗ Corresponding author. Tel.: +39 0861 266988; fax: +39 0861266989. E-mail address: (M. Perugini). 0921-4488/$ – see front matter © 2012 Elsevier B.V. All rights reserved.

properties similar to those of dioxins. They are persistent organic pollutants (POPs) (Colles et al., 2008; Liu and Liu, 2009) and, due to their carcinogenic and mutagenic effects and their endocrine disrupting action (Salihoglu and Tasdemir, 2009; Vives et al., 2008) represent a threat to human health. Since they are widespread environmental contaminants, they are found in environmental samples, in food for human consumption, in animal feed and in grassland (De Mul et al., 2008). In the environment, these compounds decompose very slowly and due to their liposolubility they accumulate in the food chain and

8 4..itat cont area2. dissolving. and it is a region where sheep extensive breeding is largely used. In assessing the impact of soil contamination on plants. GIS provides a way to include data that is not so pristine (Schuurman.38 0.31 0. milk represent the main excretion pathway for these contaminants (Schmid et al.26 represent a serious hazard for human and animal health. The parameters of lag size..05 0. all dealing with the management of data (e. Geostatistical analyses Data was analyzed using ESRI ArcGis software (Environmental Systems Research Institute.1. Dl-PCBs once absorbed by animals. modeling and presenting geographic data for a wide range of applications (Davis. these compounds accumulate in the leafy vegetation (such as grass) and in the soil.84 1.G. Italy. milk could be an ideal matrix for monitoring the PCBs presence in the environment providing exposure information through a non-invasive collection method.26 0.58 1. Total dl-PCB concentrations in soil samples are expressed as the sum of 12 congeners (WHO-TEQ ng/kg) and all data derived from the ARPA Campania web site (http://www. This permitted to estimate soil contamination values for each point where milk was sampled.g. clipping.31 0. At the same time the Agenzia Regionale Protezione Ambientale della Campania (ARPAC) monitored the levels of PCBs-dioxins in the soil.599 0. One of the chief virtues of GIS is that it allows the visualization of spatial data as well as providing a means of utilizing fuzzy data. Redlands. a regionalization. The Koenker test was performed to determine .12 0. analyzing.29 0.74 1. the Campania Region Health Authority launched a dioxin surveillance plan in order to control the pollution degree and examine contamination extent and sources.25 0.597 0.66 1. Sheep introduce dl-PCBs both eating forage which has been contaminated by atmospheric deposition and ingesting direct contaminated soil from pasture. Interpolation creates a surface. now the situation has changed. it appears that for PCBs there is an increasing concentration with decreasing soil particle size (Smith and Jones. 2004).934 0.885 0. accumulate in the liver and adipose tissues and during lactation periods.859 0. While the quantitative science prefers clear and precise facts. ingest contaminated grass and these contaminants are rapidly absorbed from the gastrointestinal tract (Faye and Sinyavskiy. CA) version 9. Once settled into soils. Number of observations 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 dl-PCBs concentrations in milk WHO-TEQ ng/kg 0. merging and intersecting) allowing to aggregate information from one layer to another. 2011). cow and buffalo) samples collection to determine the PCDD/F and dl-PCB levels. Materials and methods Sheep milk samples were collected from 26 farm located in Campania. Campania (Italy) is strongly inclined toward agriculture.07 0.. and GIS is an excellent tool for determining such characteristics (Davis. The sheep breeding was extensive livestock farming and no feed were given to the animals. Transformation. GIS offer a variety of geo-processing features. 2001). The interpolation of soil contamination values was performed to overcome the problem of different sampling locations between the variables (milk and soil). Herrera Nu˜ nez et al. obtaining a continuous map. Until the 1990s the “shepherd system” was the most important way of breeding: it was based on pastures located in places far from each other.arpacampania. The different maps obtained were compared using cross-validation and the best combination was the one with the less root mean square error. 2003.623 1.17 0.65 0. These data were obtained in collaboration with the Istituto Zooprofilattico Sperimentale del Mezzogiorno-Portici. 2001). Features on maps have spatial relationships which can be very important in many applications.16 0. 2009).75 dl-PCBs concentrations in soil WHO-TEQ ng/kg 0. The Ordinary Kriging method was used to interpolate soil contamination values. All data derive from the EU monitoring plan.3.09 0.08 0. with the “permanent” breeding. This paper aims to study the relationship between sheep’s milk and soil contaminated with dl-PCBs as a first stage of further studies to assess a valid bioindicator of environmental pollution. from point-data by geostatistical operations (Jopp et al. number of neighbors and shape of semivariogram were combined to obtain the best result. Most important for the process of integration is the interpolation of single data-points to area-wide information. Total dl-PCBs concentration in sheep’s milk samples is expressed as the sum of 12 congeners in WHO-TEQ pg/g fat in agreement with the Regulation EC 1881/2006 (Table 1). 2000).22 0. GIS is a computer-based technology and methodology for collecting.37 0. conversion or the translation of data from one unit or level to another is the next step of coupling. that acts as a natural sink. / Small Ruminant Research 106S (2012) S49–S53 Table 1 dl-PCBs concentrations in sheep milk and soil samples (WHO-TEQ-ng/kg). Residues National Plan (PNR) and Surveillance Regional Plan for PCBs and dioxins carried out in 2008 by Campania Region.10 0.57 0. 2010).4 2. dl-PCBs concentration is directly related with the environmental pollution (Zhang et al. 2008).43 0. Koenker and Global Moran’s Index statistical tests were applied..08 0..567 0.38 0. 2009. due to their lipophilic nature.S50 E. 2.163 0. Geographic Information Systems (GIS) rise as a powerful tool for data analysis.21 0. 1994). Data from different research institutes were integrated in a GIS as it allows to manage different information layers and perform spatial analyses. Since 2007. Vives et al. In this context.03 0. where dl-PCBs analyses in sheep’s milk were carried out. Esposito et al.582 0.14 0.37 0.578 6.08 0. 2.14 0. The relationship between soil and milk contaminated with dl-PCBs was analyzed using Ordinary Least Squares (OLS). according to the EPA method 1613 (USEPA. Furthermore.768 4. 2008.asp?id=10).03 0. Grazing animals as sheep. Due to atmospheric deposition.41 0.45 0.21 4.. managing. This plan was focused on milk (sheep. Esposito et al. Considering that the sheep breeding system is permanently in the open air.605 0.71 0.

If the index value is greater than 0. the set of features exhibits a clustered pattern. Therefore. When the P-value is small and the absolute value of the Z score is large enough that it falls outside of the desired confidence level. the set of features exhibits a dispersed pattern (Sullivan and Unwin.94 m and 15 neighbors (Fig. 2003). The map shows the interpolation between dl-PCBs values found in the soil and milk sampling points. The relationship between the variables presented a R2 of 0. The altitude and milk contamination were not strongly related (R2 = 0. Herrera Nu˜ nez et al. the significance of the soil contamination was determined by the robust probabilities. 1. .G. Results The best interpolation of soil contamination values was obtained using a circular semivariogram. a lag size of 4619.. Sheep’s milk data were integrated to the map obtained with Ordinary Kriging to perform regression analysis between milk and soil contaminated with dl-PCBs. for a 95 percent confidence level. the null hypothesis can be rejected. The Global Moran’s Index test was performed to the residuals of the regression to determine the existence of a spatial autocorrelation. 2002. Seddighi et al. Furthermore. the residuals were clustered. 3. / Small Ruminant Research 106S (2012) S49–S53 S51 Fig. we integrated the Digital Elevation Model (DEM) to calculate the altitude for each soil and milk sampling point in order to improve the OLS performance. If the value is less than 0. and the robust probabilities are the only reliable results (Watson and Teelucksingh. Interpolation of soil contamination values using Ordinary Kriging with a circular semivariogram. 1).05).24.18) but the altitude was strongly related to soil contamination (R2 = 0. The null hypothesis states that there is no spatial clustering of the values associated with the geographic features in the study area. heteroscedasticity. In case of statistically significance. 2000).05). The Global Moran’s Index test showed that for a 95 percent confidence level. Considering the results of the Koenker test and that heteroscedasticity could be caused from the omission of relevant regressors.53). the relationship between soil and milk contamination was statistically significant heteroscedastic (P < 0. which was statistically significant (P < 0. the linear regression model is heteroscedastic.E.

we have hypothesized that milk contamination was influenced by soil contamination in different order of magnitudes depending on altitude. confirming that the relationship between milk and soil contamination was variable within the study area. exploiting the visual advantages of GIS. Separation of data in three groups of different altitude ranges ( = 0–100 m. the altitude influences the soil contamination. Moreover. 2009). therefore. The regressions performed for each group showed R2 of 0. = 300–1000 m) in order to evaluate the influence of The visual analysis of all data integrated in GIS showed that sampling points were distributed geographically in three groups (Fig. Therefore. it was necessary to improve the OLS. Although. 2). the relationship between milk and soil contaminated with dl-PCBs . the coefficient of determination was low and the residuals were clustered (spatial autocorrelation). Performing the OLS. Considering that PCBs milk contamination is influenced by soil pollution and the latter is influenced by altitude. 0.S52 E. as the latter can influence dl-PCBs concentrations (Wang et al. which allowed the visualization on data in a geographical context. Calculating the altitude from DEM for each point. Discussion The geostatistic approach allowed the description and modeling of spatial patterns of soil contamination with dl-PCBs.35 for the groups of 0–100 m. Both. resulting in a prediction of soil pollution at unsampled locations.81 and 0. 100–300 m and 300–1000 m.. 100–300 m and 300–1000 m respectively. / Small Ruminant Research 106S (2012) S49–S53 Fig.91. Herrera Nu˜ nez et al. data was divided in the following altitude ranges: 0–100 m. = 100–300 m. The Koenker test showed that heteroscedasticity was present indicating that the regression was missing a key explanatory variable. 2. altitude and milk contamination were not strongly related. the Digital Elevation Model of the Campania Region was introduced to analyze the relationship between the variables considering altitude. altitude on the relationship between milk and soil contamination in the studied area. In addition. the resulting map was a reliable information layer to estimate pollution concentration in the locations where milk was sampled. the heteroskedasticity and autocorrelation lead to biased and inefficient estimates of parameters with considerably standard errors. The separation of data in three groups with different altitude ranges leaded to better R2 values. Thus. the separation in three groups of different altitude ranges was performed to test the hypothesis of the influence of soil in milk contamination. Kriging is the preferred method in many spatial statistical analyses and is considered the best linear unbiased estimator. The Global Moran’s Index indicated random distribution of the residuals for the three groups. 4.G.

G. C. 2011. Regul. 1212–1216. Mater. October 1994. Y. 246. L.. 2000. the results obtained in this paper. Int. R. P. Liu. A. Kotz. Herrera Nu˜ nez et al.. B. Koppen. Schmid. De Mul.. R... Gallo.. S. 5. Liu. a simple predictive model could be constructed. 278–287.J. Salihoglu.. 163. Wang. Sullivan. 2009... H.. 51. Impact of pollution on animal products.. Conclusions The approach proposed in this paper is advantageous as it allows to integrate and analyze data from different sources. Watson. E.. Fourth WHOcoordinated survey of human milk for persistent organic pollutants (POPs): Belgian results. G. Wang.. Serpe.. Colarusso. P. F. K.P. the relationship between these two variables is influenced by altitude and the R2 values decrease with higher altitude..R. L. Serpe. 1523–1531.. Jiang. Smith. New York. Zeilmaker.. John Wiley &Sons.P.. The University of the West Indies Press.. DC.. Taylor & Francis Group.. Oxford. GIS and geostatistics are essential in cutting edge research.. M..A. Since soil reflects environmental pollution very well (Zhang et al.. Altitude dependence of polychlorinated biphenyls (PCBs) and polybrominated diphenyls ethers (PBDEs) in surface soil from Tibetan Plateau. F. Esposito... 2009. Pellicanò. Lawler.. Esposito. N. .. A. 207–236. the study of environmental variables seems to be highly related to geographic location. New York. Hoboken. Chemosphere 76. Contamination levels and congener distribution of PCDDs. G. Classical and Modern..E... G.81 > 0.. P. 2009. Vives et al. Chemosphere 75. OnWord Press. Italy. Vives. A. Eur. A Practical Introduction to Econometric Methods.J. Boon. B. more data is necessary to confirm such pattern. Vinkx. thus. Unwin. G. Environ. USEPA Method 1613. S. 2003. Schoeters.. P.G.. Studer.. van Klaveren. In: Proceedings of the NATO Advanced Research Workshop on Impact of Pollution on Animal Products. Cavallo. R. Bakker. This research paper evidences the potential use of sheep milk as indicator of PCBs environmental pollution and represents an innovative method to create a monitoring contamination system. V. Off.. Beone. 2009. Dietary exposure to dioxins and dioxin like PCB’s in the Netherlands anno 2004. GIS A Short Introduction. Q. A. Modelling Complex Ecological Dynamics.91 > 0. second edition. Malisch.K... Umlauf. K. GIS: A Visual Approach. Springer. M. E. Tasdemir. Cenci. F..D. Commission Regulation (EC) No. An Introduction into Ecological Modeling. D. Temporal and local trends of PCDD/F levels in cow’s milk in Switzerland. Chemosphere 79. could be the base for future research in assessing bioindicators of environmental contamination with dl-PCB. C. Li. B.J. Li.C. 2000. M. Zhang. Serpe. W... van Leeuwen. Mariani.. V. 2009. Particles and vegetation: implications for the transfer of particle-bound organic contaminants to vegetation. Sena. 2010. G. Nonetheless. polychlorinated dibenzofurans and diodi-like polychlorinated biphenyls in cow’s milk collected in Campania. Colarusso. Gujer. R. Routledge. J. K. Nelen. Econometrics: A Practical Approach. 1186–1195. M.. Serpe. Y. Chemosphere 77. Moreover. H. Breckling. US Environmental Protection Agency (USEPA). However.... Levels and congener profiles of polychlorinated dibenzo-p-dioxins. Zhang. Sci. PCDFs and diodi-like PCBs in buffalo’s milk from Caserta province (Italy)... W. Chemosphere 73..E. Wang. China. G.. Teelucksingh. Geographic Information Analysis... Gallo. S. 164.V.. T. J. 2002. G. Union.C.. F. Hanot.. I.. R.. Washington.. References Colles. 2008.. Blackwell Publishing. G. Muller. L. 1498–1504. D’Ambrosio.. Faye.. Conflict of interest statement None of the authors have any financial or personal interest that would inappropriately influence or bias the contents of this paper.. Pharmacol. J. Iovane.A. Noel. Skejo. 129–136.. Christoph. Baldi.. A. H. P. G. Peng. Sinyavskiy. Katos.. J. X. if integrated with more data and validated with more temporal series. F. Jopp. Biot. 994–1000... Y. / Small Ruminant Research 106S (2012) S49–S53 S53 apparently weakens with altitude considering the R2 values of the three groups (0. Davis. Hazard. P. S. E. Total Environ.. Furthermore. J. Cavallo.. 2008) and sheep’s milk showed to be strongly related with soil contamination. A.L. L. Baldi. 364. Traag. M... Reuter.. Li. G. Distribution of polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDDs/Fs) and dioxin-like polychlorinated biphenyls (dioxin-like PCBs) in the soil in a typical area of eastern China. 2009. Seddighi. 2001.A. M. Neugebauer... 959–966... Tetra-through Octa-chlorinated Dioxins and Furans by Isotope Dilution HRGC/HRMS. 2003.. 907–914.. 2008.35).K. 5–24.. Jones. Ding.O. S. and industrial city in Turkey.I.. Y. 1881/2006 (19 December 2006) setting maximum levels for certain contaminants in foodstuffs. L. Zenneg. P. Dewolf. Toxicol. Zhang.. 2008. P. 341–348. 2008.E... X.. Hazard. D. 2004. Schuurman. PCDD/PCDF pollution in soils and sediments from the Pearl River Delta of China.. Levels of PCDD/Fs and trace elements in superficial soils of Pavia Province (Italy). Hoogenboom. Mater. The results obtained in this study evidence that milk and soil contaminated with dl-PCBs are strongly correlated suggesting that the sheep milk reflects environmental PCBs contamination. 34 (7). Huang. Chemosphere 53. Prediction of the PCB pollution in the soils of Bursa. W.. Guarino. USEPA Office of Water Engineering and Analysis Division.