Environment International 77 (2015) 106115

Contents lists available at ScienceDirect

Environment International
journal homepage: www.elsevier.com/locate/envint

Full length article

The effects of ooding on dioxin and PCB levels in food produced on

industrial river catchments
Iain R. Lake a,, Christopher D. Foxall a, Alwyn Fernandes b, Mervyn Lewis c, Martin Rose b, Oliver White c,
Andrew A. Lovett a, Shaun White b, Alan Dowding d, David Mortimer d
a

School of Environmental Sciences, University of East Anglia, Norwich NR4 7TJ, UK

The Food and Environment Research Agency (FERA), Sand Hutton, York YO41 1LZ, UK
Askham Bryan College, Askham Bryan, York YO23 3FR, UK
d
Food Standards Agency, Aviation House, 125 Kingsway, London WC2B 6NH, UK
b
c

a r t i c l e

i n f o

Article history:
Received 14 July 2014
Received in revised form 15 January 2015
Accepted 15 January 2015
Available online 7 February 2015
Keywords:
Dioxins
PCBs
Flooding
Beef
Milk
Lamb
Food
Risk assessment
Soil
Grass

a b s t r a c t
This research examined the effect of regular ooding upon PCDD/F and PCB levels in milk, beef and lamb, produced on the oodplains of industrial river catchments. Our unique dataset included more than 200 samples
analysed for PCDD/Fs and PCBs over two data collection phases (19981999 & 20082010) from working
farms. A robust paired study design was adopted with samples taken from ood-prone farms and nearby control
farms not subject to ooding.
On industrial river catchments regular ooding is associated with higher PCDD/F and PCB levels in soils and grass.
This contamination may be transferred to food but the impact varied by food type. These contrasts may be due to
physiological differences between animals, the ages at which they are sent to market and differences in animal
husbandry.
To minimise the risks of producing food on ood-prone land in industrial river catchments, as well as on any land
with elevated PCDD/F and PCB levels, this research suggests a number of options. The choice of livestock may be
important and as an example in our study beef cattle accumulated PCDD/Fs to a higher degree than sheep. Land
management may also play a role and could include minimising the time that livestock spend on such land or
feeding commercial feed, low in PCDD/Fs and PCBs, where appropriate.
2015 Published by Elsevier Ltd.

1. Introduction
Polychlorinated dibenzo-p-dioxins and furans (dioxins; PCDD/Fs)
and polychlorinated biphenyls (PCBs) are recognised persistent organic
pollutants (POPs) and are reported to have negative effects on human
health (Schecter and Gasiewicz, 2003). In the past, PCBs were commonly used in many industrial applications, most notably as insulating and
cooling uids for industrial transformers and capacitors. Conversely
PCDD/Fs have never been produced intentionally, but are typically
formed through combustion activities or as a by-product of the production of organo-chlorine chemicals. These compounds are now ubiquitously present in the environment resulting in concerns relating to
their environmental persistence, their bioaccumulation potential, and
their toxicity. Health impacts associated with these compounds include
immunotoxicity, carcinogenicity and developmental and reproductive
effects (Schecter and Gasiewicz, 2003).
In more developed countries, industrial combustion emissions are
tightly regulated and PCB production has ceased. Consequently, levels
Corresponding author.
E-mail address: I.Lake@uea.ac.uk (I.R. Lake).

http://dx.doi.org/10.1016/j.envint.2015.01.006
0160-4120/ 2015 Published by Elsevier Ltd.

of PCDD/Fs and PCBs in the environment (Alcock and Jones, 1996;

Schuster et al., 2011), food (Durand et al., 2008; Harrison et al., 1998)
and humans (Alivernini et al., 2011; Lignell et al., 2009) are declining.
Therefore, the issue is one of managing a historical legacy, complicated
by the environmental persistence (congener half-lives measured in
decades) of many of these compounds. In less developed countries,
the situation is different, and rapid industrialisation has led to increased
combustion emissions of PCDD/Fs, which are only starting to be regulated (Zhao et al., 2011). The high levels of PCBs in these settings may be
due to leaks and spills, illegal disposal, incineration, historical contamination (Xia et al., 2012) or due to the practise of importing waste from
more developed regions. In some of these countries, PCDD/F and PCB
levels in human tissue are now similar to those in European populations
(e.g. China; Shen et al., 2009). In addition PCDD/F and PCB concentrations are still rising in some populations (e.g. China; Sun et al., 2011).
Understanding the mechanisms through which PCDD/Fs and PCBs
can be transferred from the environment to humans is essential to minimise human exposure. The major route of transfer is through food and
around 90% of human intake of PCDD/Fs and PCBs is through this route
(Liem et al., 2000). These compounds enter the food chain through several pathways. Atmospheric deposition over crops for food or animal

I.R. Lake et al. / Environment International 77 (2015) 106115

feed is a recognised pathway for terrestrial food sources. In river systems where contamination has occurred (typically those with an industrial history), river sediment can act as a sink for PCDD/Fs and PCBs
providing a long term source of release for these pollutants (Fattore
et al., 2002; Zhang et al., 2008). These contaminants can then be transferred to agricultural land through river ooding. This may become increasingly important as climate change leads to increased ooding in
some parts of the world, mobilising historically contaminated sediments containing PCDD/Fs and PCBs (Boxall et al., 2009; Hilscherova
et al., 2007).
Most of the published work on the impact of river ooding upon
PCDD/F and PCB levels in food has been conducted by this research
group. Other published papers (e.g. Hendriks et al., 1996) have tended
to be more restricted in scope and have examined smaller numbers
and types of samples. A major study in the United States focused on
the impact of ooding in a river basin in Michigan. The sediments had
become contaminated with dioxins due to industrial activities dating
to the early 1900s. The primary focuses of the study were on the impact
to soils (Demond et al., 2008) and serum (Hedgeman et al., 2009), but
not on food.
In 2005 we reported that soils on ood-prone land in industrial river
catchments had higher levels of PCDD/Fs and PCBs and that milk produced on ood-prone land had higher levels of PCDD/Fs and PCBs
than that produced on other land (Lake et al., 2005). In 2014 we reported that such contrasts are also evident in beef produced on ood-prone
land (Lake et al., 2014). However, to date we have published only some
of the data collected during these projects (Foxall et al., 2004; Lake et al.,
2012), and by focussing upon individual food types (cows' milk and
beef) several important research questions have not been addressed.
In this paper we pool data from two major data collection phases
(19981999 and 20082010) for cows' milk, and combine it with additional data for meat from beef cattle and sheep. This consists of over 200
samples covering environmental characteristics (soil and grass), commercial animal feed and food (milk, beef and lamb). The major research
questions that these data permit us to address are:
1. What impact does regular river ooding have upon environmental
contamination (soil and grass) in industrial river catchments?
2. To what extent do any impacts transfer into the food chain, and how
does this differ between food types (milk, beef and lamb)?
3. If environmental contamination is transferred to food then how is
this changing over time?
The answers to these three questions will enable a fourth to be
addressed.
4. What are the implications for the risk management of producing
food on the ood plains of industrial river catchments?

2. Materials and methods

Four river systems in central UK were examined. The rst three were
(i) the Doe Lea/Rother/Don (which together form a single river system),
(ii) the Trent and (iii) the Aire/Ouse (see Fig. 1) which ow through
substantial urban and industrial areas. Historically these catchments
contained a number of coking and chemical manufacturing plants and
chlorinated chemical facilities. They also ow through several major
cities. The River Dee (iv) was chosen as a comparator due to the lack
of an industrial history within its catchment. Sampling was conducted
on these river systems in 2 phases (phase 1: 19981999 & phase 2:
20082010).
Maps of farm distributions and ood history along the length of the
river systems were obtained and used to identify farms where a significant proportion of the land was subject to regular ooding. For each
ood-prone farm, a nearby farm whose land was not subject to ooding
was selected as a control. Control farms were selected in areas where

107

they would be expected to use similar farming practices and to be subject to similar levels of aerial deposition of PCDD/Fs and PCBs as the
neighbouring ood-prone farms (e.g. comparable proximity to industrial facilities and major roads). The application of sewage sludge to land
can be a source of PCDD/Fs and PCBs, but none of these selected farms
had received any sewage sludge in the 20 years before sampling. The
comparison of PCDD/Fs and PCB levels between these pairs of farms
should provide the strongest evidence with which to examine the impact of regular ooding events. Table 1 presents an overview of the
farm pairs analysed, subdivided by livestock product and date. All livestock products were collected in the autumn/early winter when the animals would have been grazing outdoors for an extended period and at a
point when they would typically be sent to market. Milk samples were
taken from the bulk tank on each farm and so represent the mean production for the herd. Lamb and beef samples were taken from two animals on each farm and the results averaged to produce a single value.
Extended details of the sampling procedures are presented elsewhere
(Lake et al., 2005, 2014).
To provide support for any trends emerging in the milk, beef and
lamb data, samples were collected of dietary inputs to the livestock.
These consisted of grass and soil which are consumed inadvertently
by livestock whilst foraging, and commercial animal feed which is fed
to dairy cattle throughout the year (Lake et al., 2005). Beef cattle are
also fed commercial feed whilst they are indoors. However, our samples
were taken at the end of the summer when the animals were still outdoors or had just moved indoors, and would have not been fed commercial feed for over 6 months. Therefore, such feed is unlikely to have
much inuence on contamination levels (Lake et al., 2014) because
the animals would not have been fed this over the whole of the summer.
Soil and grass samples were collected from elds regularly grazed by the
herd using previous published methods (Lake et al., 2005). In the case of
ood-prone farms, samples were taken from locations on the farm with
a history of regular ooding (most farms ooded at least once a year).
This means that we are examining the effect of regular ooding not a
specic event. Food samples from all farm pairs (53 in total), were
analysed for PCDD/Fs and PCBs. Due to resource constraints the majority
of the soil samples (41 of 47 pairs) but around half of the grass samples
(23 of 53 pairs) were analysed for PCDD/Fs and PCBs.
PCDD/F and PCB concentrations were determined using validated
methods that were accredited to the ISO 17025 standards, and are
described in detail elsewhere (Fernandes et al., 2004). In brief, representative sample aliquots were fortied with 13C-labelled analogues
of target compounds and exhaustively extracted using mixed organic solvents. Extracts were fractionated on activated carbon, yielding
planar (PCDD/Fs, laterally substituted PCBs) and non-planar fractions (ortho-PCBs) which were chromatographically puried on activated basic alumina. Analytical measurement for the PCDD/Fs and
laterally substituted PCBs was carried out using high resolution gas
chromatography coupled to high resolution mass spectrometry.
Method control was provided by the inclusion of a method blank
and reference material which were included with every set of samples. The methodology was continuously validated by successful participation in inter-comparison studies (e.g. Norwegian Institute of
Public Health, 2009) over the duration of the research. The full quality control and assurance measures undertaken for all analytes are
reported elsewhere (Lake et al., 2005).
The analyses all included the seventeen 2,3,7,8-Cl substituted PCDD/
F congeners, four non-ortho substituted PCBs (IUPAC numbers: 77, 81,
126, and 169) and twenty-one ortho substituted PCB congeners
(IUPAC numbers: 18, 28, 31, 47, 49, 51, 52, 99, 101, 105, 114, 118, 123,
128, 138, 153, 156, 157, 167, 180, and 189). WHO-1998 TEFs (van den
Berg et al., 1998) were used rather than the more recent WHO-2005
TEFs to enable comparison with data previously reported. In this
paper we report summed TEQ values (incorporating PCDD/F, ortho
and non-ortho PCB concentrations), PCDD/F TEQ is also reported, alongside the International Council for the Exploration of the Seas-6 Indicator

108

I.R. Lake et al. / Environment International 77 (2015) 106115

Fig. 1. The sampling areas and catchments of the Dee, Trent, the Doe Lea/Rother/Don and the Aire/Ouse.

PCBs (ICES6) PCB concentrations (IUPAC numbers: 28, 52, 101, 138, 153,
and 180). These 3 were chosen because they are widely reported in the
literature and also form part of the regulatory limits for PCDD/Fs and
PCBs in food (e.g. European Union; EU, 2012).
All concentrations are reported as upper bound (i.e. where congeners
were not detected, they were assumed to be present at the LOD). The
beef, lamb and milk data are presented on a fat weight basis (using either

the Werner-Schmidt or Rose-Gottlieb method). Soil and grass are presented on a dry weight basis and feed on a whole weight basis.
As part of the analysis PCDD/F and PCB concentrations in 19981999
were compared to levels in 20082010. However, throughout the
study samples were analysed in the same laboratory, using essentially the same method of extraction and purication, and instrumental
analysis (using high-resolution GC coupled to high resolution mass

Table 1
Study overview by livestock product and date.

Milk; no. ood-prone/control pairs

Beef; no. ood-prone/control pairs
Lamb; no. ood-prone/control pairs

19981999 non-industrial river

19981999 industrial rivers

20082010 industrial rivers

Dee

Doe Lea/Rother/Don

Trent

Trent

Aire/Ouse

10

13

5
5
5

5
1

I.R. Lake et al. / Environment International 77 (2015) 106115

spectrometry HRGCHRMS), over the duration of the project. HRMS

parameters, particularly selectivity and dynamic sensitivity have changed
little over this period (instrument geometry and design have remained
stable), and it has been possible to maintain the limits of detection for individual PCDD/F and PCB congeners (typically 0.01 ng/kg for PCDD/Fs and
non-ortho PCBs). This is reected in the individual TEQs for PCDD/Fs, nonortho PCBs and ortho-substituted PCBs which have also remained at
~0.03 ng/kg, 0.01 ng/kg and 0.02 ng/kg respectively. Therefore, we are
condent that differences over time reported in this paper are not due
to varying laboratory techniques or limits of detection.
These data were analysed using statistical analysis as appropriate.
Where there were a large number of samples distributed across the
sampling programme (e.g. soil, milk), multiple regression analysis was
undertaken examining the inuence of multiple factors (e.g. oodprone/control location and date) upon PCDD/F and PCB levels. Factors
were only included in the statistical analysis if there were at least
20 samples for each variable of interest (e.g. 20 samples in the 1998
1999 sampling period). Where sample numbers were smaller (e.g.
grass, commercial feed), the statistical analysis focussed upon comparing two groups of samples (e.g. comparing contaminant levels in feed
between 20082010 and 19981999) and was only undertaken if
there were at least 10 samples in each group. For these analyses,
MannWhitney U tests were used if the samples were not paired; otherwise the Wilcoxon signed-ranks test was used. Non-parametric tests
were used due to their enhanced robustness and the general nonnormal distributions of these data. Throughout the statistical analysis,
the relatively small numbers of samples analysed in this study, implies
that very pronounced differences would be required (e.g. between
ood-prone and control sites) to achieve statistical signicance.
3. Results
3.1. What impact does regular river ooding have upon environmental
contamination in industrial river catchments?
The summed TEQ, PCDD/F TEQ and ICES6 results for the soil and
grass samples are presented in Table 2 subdivided by river system and
sampling period. This indicates that on the industrial river systems
(Trent, Aire/Ouse, Doe Lea/Rother/Don) median soil summed TEQ,
PCDD/F TEQ and ICES6 were higher in ood-prone soils. These contrasts
were greatest on the Trent. A more robust analysis compares the levels
between matched pairs of ood-prone and control farms. A line chart
comparing levels between industrial ood-prone and control pairs is illustrated in Fig. 2ac and shows that for the majority of farm pairs,
levels in soil were higher in ood prone samples (29/35 TEQ, 29/35
PCDD/F TEQ and 32/35 ICES6). In contrast Table 2 and Fig. 2df indicate
that on the Dee (non-industrial), summed TEQ, PCDD/F TEQ and ICES6

109

levels are lower, and hence the contrasts between ood-prone and control sites are less marked than on industrial river systems. On three
farms on industrial river systems the impact of ooding was examined
in more detail by comparing two samples of soil simultaneously collected from the same eld. One soil sample was collected from a part of the
eld prone to ooding and another from a location in the same eld
which was not prone to ooding. In all cases summed TEQ, PCDD/F
TEQ and ICES6 were over four times higher on the soil from the oodprone part of the eld (Supplementary material 1).
The soil data were examined statistically using Wilcoxon signedranks tests which indicated that summed TEQ, PCDD/F TEQ and ICES6
were signicantly higher on ood-prone samples from industrial river
systems (p b 0.001). This analysis was taken further using multivariate
regression. The results are listed in Supplementary material 2, and indicated that ood-prone soils have signicantly higher summed TEQ,
PCDD/F TEQ and ICES6 concentrations, but that levels found on oodprone soils on the non-industrial river Dee are similar to levels found
in control soils.
Grass samples were taken from the same locations as the soil samples. A subset were analysed for PCDD/Fs and PCBs. There were weak
positive correlations between summed TEQ, PCDD/F TEQ and ICES6 in
the grass sample in comparison to the corresponding soil sample from
the same location (Supplementary material 3). In spite of these weak
correlations Table 2 indicates that PCDD/F and PCB levels are generally
higher in grass from ood-prone sites on industrial river systems. A
comparison of the differences between ood-prone and control pairs
(Fig. 2gi) indicated that a majority were higher on industrial river systems (16/22 summed TEQ, 16/22 PCDD/F TEQ and 14/22 ICES6). Taking
all these grass data from industrial river systems, a Wilcoxon signedranks test indicated that levels were signicantly higher in grass on
ood-prone sites (summed TEQ, p = 0.034; PCDD/F TEQ, p = 0.040;
ICES6, p = 0.017) in comparison to control sites. Only two grass samples
(1 pair) were analysed from the non-industrial river Dee precluding statistical analysis. Analysis of this pair showed that levels were very low
on both samples (Fig. 2jl). In common with the soil data, three sites
were identied where grass samples were taken from ood-prone and
non-ood prone parts of the same eld. Grass levels were greater on
the ood-prone part of the eld in 2 from 3 farm pairs (Supplementary
material 1).
3.2. To what extent do any impacts transfer into the food chain, and how
does this differ between food types?
Flooding appears to be a mechanism for the transfer of PCDD/Fs and
PCBs from rivers to soil and grass on the ood plains of industrial river
catchments, but are these contaminants transferred to food produced
on these sites? The summed TEQ, PCDD/F TEQ and ICES6 results for

Table 2
Comparisons of median soil and grass summed TEQa, PCDD/F TEQa and ICES6b across river systems and sampling periods.
19981999 non-industrial
river

19981999 industrial rivers

Dee

Doe Lea/Rother/Don

20082010 industrial rivers

Trent

Trent

Aire/Ouse

Summed
TEQ

PCDD/F
TEQ

ICES6

Summed
TEQ

PCDD/F
TEQ

ICES6

Summed
TEQ

PCDD/F
TEQ

ICES6

Summed
TEQ

PCDD/F
TEQ

ICES6

Summed
TEQ

PCDD/F
TEQ

ICES6

Soil
Flood prone
Control
Overall

4.036
4.176
4.1712

3.556
3.656
3.6512

1.846
0.906
1.0512

12.815
10.755
11.4110

12.045
10.225
10.5010

1.485
1.015
1.2110

21.6210
4.3910
7.6020

19.1710
4.0310
7.0420

3.1910
0.9010
1.6120

22.3414
5.0614
8.4828

17.3614
4.5814
7.8028

32.4714
0.9514
1.4828

8.596
4.466
5.8712

8.076
4.076
5.2512

3.956
0.836
1.1912

Grass
Flood prone
Control
Overall

0.121
0.111
0.122

0.081
0.071
0.082

0.231
0.191
0.212

1.311
0.301
0.812

0.991
0.151
0.572

0.901
0.901
0.902

0.253
0.293
0.286

0.173
0.173
0.176

0.273
0.333
0.306

0.6112
0.2312
0.2724

0.4512
0.1512
0.2024

1.3712
0.2212
0.2824

0.876
1.206
1.2012

0.706
0.946
0.9412

0.436
0.406
0.4112

Figures in subscript are sample sizes.

a
ng TEQ/kg dry wt.
b
ug/kg dry wt.

I.R. Lake et al. / Environment International 77 (2015) 106115

(b) Soil PCDD/F TEQ (industrial rivers)

1000
PCDD/F TEQ (ng/kg dry)

Summed TEQ (ng/kg) dry)

(a) Soil TEQ (industrial rivers)

1000

100

10

100

10

0.1

0.1

0.1

0.01
Flood-prone pair

Control pair

0.1

0.01
Flood-prone pair

0.1
Flood-prone pair
Control pair
(i) Grass ICES6 (industrial rivers)
10

0.1
Flood-prone pair
Control pair
(l) Grass ICES6 (non-industrial rivers)
1
ICES6 (ug/kg dry)

PCDD/F TEQ (ng/kg dry)

Summed TEQ (ng/kg) dry

0.01
0.01
Flood-prone pair
Control pair
Flood-prone pair
Control pair
(j) Grass TEQ (non-industrial rivers)
(k) Grass PCDD/F TEQ (non-industrial
rivers)
1
1

10

10

ICES6 (ug/kg dry)

PCDD/F TEQ (ng/kg dry

Summed TEQ (ng/kg dry)

1
1
Flood-prone pair
Control pair
Flood-prone pair
Control pair
(g) Grass TEQ (industrial rivers)
(h) Grass PCDD/F TEQ (industrial
rivers)
10
10

100

0.1
Flood-prone pair
Control pair
(f) Soil ICES6 (non-industrial rivers)

ICES6 (ug/kg dry)

PCDD/F TEQ (ng/kg dry)

Summed TEQ (ng/kg dry)

1
1
Flood-prone pair
Control pair
Flood-prone pair
Control pair
(d) Soil TEQ (non-industrial rivers)
(e) Soil PCDD/F TEQ (non-industrial
rivers)
10
10

(c) Soil ICES6 (industrial rivers)

1000
ICES6 (ug/kg dry)

110

Control pair

0.1
Flood-prone pair

Control pair

Fig. 2. Comparison of summed TEQ, PCDD/F TEQ and ICES6 for matched pairs of ood-prone and control farms for soil and grass.

the food samples (milk, beef and lamb) are presented in Table 3
subdivided by river system and sampling period. A more robust analysis
compares the levels between matched pairs of ood-prone and control
farms. A line chart comparing levels between ood-prone and control
farms is illustrated in Fig. 3 for industrial river systems with different
food samples and sampling periods.
In an earlier paper (Lake et al., 2005) we noted that during 1998
1999, milk from ood-prone farms on industrial river systems had signicantly higher summed TEQ and PCB (ICES7) concentrations. This effect was also seen in summed TEQ for beef produced on ood-prone
farms during 20082010 (Lake et al., 2014), and Table 3 and Fig. 3gi indicate that this also appears to be the case for PCDD/F TEQ and ICES6.
However, for milk produced on the Trent in 20082010 only 3/5 farm
pairs had higher summed TEQ and ICES6 on the ood-prone farm.

Only 2/5 ood-prone farm pairs had higher PCDD/F TEQ. Analysis of
the lamb data (Fig. 3jl), showed that 3/6 ood-prone farm pairs had
higher summed TEQ, PCDD/F TEQ and ICES6 during 20082010. The
common feature with both these examples is the lower overall levels
of PCDD/Fs and PCBs in comparison to beef (20082010) and milk
(19981999). Therefore, should a difference exist a large number of
farm pairs would need to be analysed for any signicant difference between ood-prone and control farms to emerge.
To explain contrasts between ood-prone and control farms, it is
important to consider whether commercial feed may play a role. Little
commercial feed is fed to beef cattle and lamb during the outdoor
period. For dairy cows, commercial feed is more important because it
is provided throughout the year. Commercial feed is distributed on a regional basis and so unlikely to explain contrasts between ood-prone

I.R. Lake et al. / Environment International 77 (2015) 106115

111

Table 3
Comparisons of median food summed TEQa, PCDD/F TEQa and ICES6b across river systems and sampling periods.
19981999

Flood prone
Control
Overall

20082010

Non-Industrial River

Industrial Rivers

Dee

Doe Lea/Rother/Don

Industrial Rivers
Trent

Trent

Aire/Ouse

Summed
TEQ

PCDD/F
TEQ

ICES6

Summed
TEQ

PCDD/F
TEQ

ICES6

Summed
TEQ

PCDD/F
TEQ

ICES6

Summed
TEQ

PCDD/F
TEQ

ICES6

Milk
1.559
1.309
1.3218

0.679
0.669
0.6718

2.159
1.639
2.0818

4.9210
2.9110
3.9120

2.5710
1.1710
1.8710

4.5810
3.6610
4.1220

3.5513
2.1813
2.8626

2.1213
1.1013
1.6126

3.4213
2.4113
2.9226

1.365
0.945
0.9710
Beef
2.905
2.405
2.4210
Lamb
1.265
1.065
1.0810

0.725
0.455
0.5410

1.595
1.175
1.2810

1.125
1.015
1.1110
0.655
0.585
0.6010

Flood prone
Control
Overall
Flood prone
Control
Overall

Summed
TEQ

PCDD/F
TEQ

ICES6

7.555
2.555
3.9910

4.075
3.685
3.7410

1.805
1.565
1.6910

7.065
2.705
6.1710

6.015
4.885
4.8810

0.481
0.691
0.592

0.251
0.351
0.302

0.691
1.071
0.882

Notes.
Figures in subscript are sample sizes.
a
ng TEQ/kg fat wt.
b
ug/kg fat wt.

and control farms. It is also tightly regulated for PCDD/Fs and PCBs (EU,
2002, 2006). Thirty commercial feed samples were taken as part of the
sampling design. These were characterised by low levels of PCDD/Fs
and PCBs, and MannWhitney U tests indicated no signicant differences
in summed TEQ, PCDD/F TEQ and ICES6 between feeds from ood-prone
and control farms (summed TEQ p = 0.629; PCDD/F TEQ p = 0.877;
ICES6 p = 0.846; n = 30).
One unique features of this study is that samples of soil and grass
were taken concurrently from each farm. This permits us to consider
how the levels of PCBs and PCDD/Fs in food are affected by the type of
farming undertaken. Fig. 4 plots the summed TEQ, PCDD/F TEQ and
ICES6 concentration in the environment on each farm (soil and grass
levels) against the food concentration (milk, beef and lamb). For greater
comparability these graphs focus upon 20082010. Fig. 4a and b indicates that for a given summed TEQ in grass and soil on a farm, the
summed TEQ in beef from such farms is higher than the summed TEQ
observed in lamb or milk (i.e. most of the beef data points are above
those for lamb and milk). Fig. 4c and d demonstrate that this trend is
also apparent for PCDD/F TEQ. Fig. 4e and f indicate that ICES6 shows
a different pattern, and given an ICES6 level in grass and soil on a
farm, the ICES6 level in beef from such farms is similar to that found
in lamb but still higher that that observed in milk. Further evidence
for the differing effects of varying food systems comes from three
farms on the River Trent which had milk as well as beef cattle. On all 3
farms levels of summed TEQ, PCDD/F TEQ and ICES6 were higher in
beef than in milk (Supplementary material 4). In Fig. 4, the lack of a
strong association between PCB and PCDD/F levels in the environment
and food is unsurprising as levels in soil and grass are likely to vary
hugely across a farm. Given these uncertainties it is notable that large
differences between animal products still emerge.
3.3. If environmental contamination is transferred to food then how is this
changing over time?
Elevated PCDD/F and PCB levels are present in soils and grass on
ood-prone farms and a transfer of these to food is apparent in some
food products (beef 20082010 and milk 19981999). In terms of
trends over time, Table 2 indicates that a comparison between soil samples from the River Trent in 19981999 and 20082010 presents little
evidence of a reduction in summed TEQ, PCDD/F TEQ or ICES6. This is
conrmed in the multivariate statistical analysis presented in Supplementary material 2 where the non-signicance of the variable
representing the 20082010 sampling period indicates that no

signicant change in summed TEQ, PCDD/F TEQ or ICES6 has occurred

in soils in the decade between sampling periods. A detailed statistical
analysis of PCDD/F and PCB levels on grass over time is not possible
due to the small numbers of grass samples analysed on the River
Trent in 19981999 (n = 6). However, the data presented in Table 2
do not provide any indication of declining levels.
Therefore the level of environmental contamination is similar
between 19981999 and 20092010. However, when the food data is
examined (Table 3) there is a 66% drop in PCDD/F and PCB levels in
milk between 19981999 and 20082010. MannWhitney U tests indicated that these reductions are statistically signicant (summed TEQ
and PCDD/F TEQ p b 0.001; ICES6 p b 0.01; n = 36). Further evidence
for this decline comes from one of the unique features of our data, two
milk farms sampled once in 19981999 and again in 20082010.
These data are presented in Supplementary material 5 and indicate
that summed TEQ, PCDD/F TEQ and ICES6 have all declined in milk
over this period. One explanation for this reduction may be lower levels
of PCDD/Fs and PCBs in commercial feed. For commercial feed (n = 29;
data not shown) between 19981999 and 20082010 median summed
TEQ declined marginally (0.07 vs 0.06 ng TEQ/kg whole, MannWhitney U p = 0.01), PCDD/F TEQ declined (0.04 vs 0.03 ng TEQ/kg whole,
p = 0.002) and ICES6 also declined (0.12 vs 0.06 g/kg whole wt, p b
0.01). Again we highlight our statement expressing our condence
that differences over time reported in this paper are not due to varying
laboratory techniques or limits of detection.
4. Discussion
This paper has presented the most comprehensive study of the potential impacts of regular river ooding events on PCDD/F and PCB
levels in food to date. Its robustness is enhanced through the use of a
paired study design to minimise other potential differences between
ood-prone and control farms. Throughout the study, identical sampling protocols and sample processing, at the same laboratory using
identical analytical methods, minimise variability in the data from
these sources. We have also argued that our method largely eliminates
the possibility that changing concentrations over time are due to changing laboratory practices. The relevance of this study for risk assessment
and potentially for risk management is enhanced by its use of data from
working farms.
Using 41 pairs of soil samples from ood-prone and control farms
across 4 river systems (River Trent, Doe Lea/Rother Don, Aire and
Dee) and two time periods (19981999 and 20082010) in central

I.R. Lake et al. / Environment International 77 (2015) 106115

(a) Milk 1998-1999 summed TEQ

(b) Milk 1998-1999 PCDD/F TEQ

8
6
4
2

Control pair

(d) Milk 2008-2010 summed TEQ

Control pair

Control pair

6
4
2

Control pair

(j) Lamb 2008-2010 summed TEQ

6
4
2

0.5

Control pair

Control pair

6
5
4
3
2
1

Control pair

0
Flood-prone pair
12

10

6
4
2

Control pair

(i) Beef 1998-1999 ICES6

10

8
6
4
2

Control pair

(k) Lamb 2008-2010 PCDD/F TEQ

Dioxin TEQ (ng/kg fat)

(f) Milk 2008-2010 ICES6

0
Flood-prone pair

Control pair

(l) Lamb 2008-2010 ICES6

1.5

0
Flood-prone pair

(h) Beef 1998-1999 PCDD/F TEQ

Dioxin TEQ (ng/kg fat)

0
Flood-prone pair

10

0
Flood-prone pair

ICES6 (ug/kg fat)

10
Summed TEQ (ng/kg fat)

ICES6 (ug/kg fat)

(g) Beef 1998-1999 summed TEQ

Summed TEQ (ng/kg fat)

(e) Milk 2008-2010 PCDD/F TEQ

Dioxin TEQ (ng/kg fat)

Summed TEQ (ng/kg fat)

0
Flood-prone pair

12

10

0
Flood-prone pair

10

0
Flood-prone pair

12

0
Flood-prone pair

8
ICES6 (ug/kg fat)

10

0
Flood-prone pair

(c) Milk 1998-1999 ICES6

12
Dioxin TEQ (ng/kg fat)

Summed TEQ (ng/kg fat)

12

14
12

1.5

ICES6 (ug/kg fat)

112

1
0.5

0
Flood-prone pair

10
8
6
4
2

Control pair

0
Flood-prone pair

Control pair

Fig. 3. Comparison of summed TEQ, PCDD/F TEQ and ICES6 for matched pairs of ood-prone and control farms on industrial river catchments for milk, beef and lamb.

England we indicate that summed TEQ, PCDD/F TEQ and ICES6 levels are
signicantly higher in ood-prone farms. This result is strengthened by
the observation that no such contrast exists on the River Dee, whose
catchment is not industrial. In addition 23 pairs of grass samples were
collected at the same locations and indicated signicantly higher
summed TEQ, PCDD/F TEQ and ICES6 levels for ood-prone sites. The
magnitude of effect was larger for soil than for grass which is to be
expected given the known persistence of PCDD/Fs and PCBs in soil and
also that levels of such compounds on grass will be affected by shortterm factors such as rate of growth, air temperature and precipitation.
Given these factors it is notable that contrasts in grass were still evident
and statistically signicant. We believe that this is the most comprehensive study to date indicating that river ooding on industrial river catchments can lead to elevated PCDD/F and PCB levels in soils and grass.

To put the levels of environmental contamination found into context, the soil results were compared to those reported in previous UK
studies. These divide soils into those from urban and rural locations,
and the two main studies (DEFRA, 2002; Environment Agency, 2007)
report mean summed TEQ in soil from urban locations of between
28.0 and 11.1 ng TEQ/kg dry respectively. These values compare
favourably to the levels found on our control sites on industrial river
catchments. These sites do not ood, but have a history of
industrialisation, so the elevated levels are likely to be due to atmospheric deposition. This study indicates that on such industrial river
catchments, levels are increased further in ood-prone locations, a likely consequence of deposited river sediment that is relatively high in
PCDD/Fs and PCBs. On our non-industrial river system, PCDD/F and
PCB levels from both our ood-prone and control sites were not

I.R. Lake et al. / Environment International 77 (2015) 106115

(a) Summed TEQ level in soil vs food

(b) Summed TEQ level in grass vs food

10

Beef
Lamb
Milk

Food summed TEQ (ng/kg fat)

Food summed TEQ (ng/kg fat)

10

6
4
2
0

10

20
30
40
50
Soil summed TEQ (ng/kg) dry

6
4
2

0.5
1
1.5
2
2.5
Grass summed TEQ (ng/kg) dry

Beef
Lamb
Milk

Food Dioxin TEQ (ng/kg fat)

Food Dioxin TEQ (ng/kg fat)

(d) PCDD/F TEQ level in grass vs food

3
2
1
0
20
40
Soil PCDD/F TEQ (ng kg) dry

Beef
Lamb
Milk

4
3
2
1
0

60

0.5
1
1.5
2
Grass PCDD/F TEQ (ng/kg) dry

2.5

(f) ICES6 level in grass vs food

15

Beef
Lamb
Milk

10

15

Food ICES6 (ug/kg fat)

(e) ICES6 level in soil vs food

Food ICES6 (ug/kg fat)

Beef
Lamb
Milk

60

(c) PCDD/F TEQ level in soil vs food

113

Beef
Lamb
Milk

10

0
0

50

100

150

200

Soil ICES6 (ug/kg) dry

2
3
Grass ICES6 (ug/kg) dry

Fig. 4. Comparing summed TEQ, PCDD/F TEQ and ICES6 in the environment versus food.

statistically different. Both compare well with the data reported in the
two main previous studies from rural locations ((2.5 ng TEQ/kg dry;
DEFRA, 2002) (5.2 ng TEQ/kg dry; Environment Agency, 2007)).
However, the extent to which these levels are transferred to food
produced on the ood plains of industrial river catchments depends
upon the farming activity undertaken and husbandry practices. For example, an elevation in summed TEQ and PCDD/F TEQ was observed on
ood-prone farms producing beef but not lamb. The main reason for
this difference may be that summed TEQ and PCDD/F TEQ levels were
much lower in lamb than in beef. Consequently small differences in
levels between ood-prone and control farms are unlikely to be detected in lamb. Levels of ICES6 in lamb are comparable to those in beef and
so the lack of contrast between ood-prone and control sites may be
due to the low number of lamb pairs. There are a number of reasons
why beef cattle and sheep may exhibit contrasting levels of PCDD/Fs
and PCBs in their meat on land with similar levels of contamination.
Possible causes could include physiological differences between sheep
and beef affecting the amount of PCDD/F and TEQ that are ingested
and retained. In addition it has been found that ruminants consume
notable quantities of soil (Healey, 1972) which are higher in PCDD/Fs
and PCBs in comparison to grass. Therefore, differing quantities of soil

ingestion may affect the body burden of PCDD/F and PCBs. Contrasts
may also be related to the different ages of the animals at slaughter
(18 vs 4 1/2 months for cattle vs sheep) as there is evidence that the
body burden of PCDD/F and PCB increases with age in some mammals
(Lorber et al., 1997). Commercial feed is an unlikely explanation as
these animals are rarely fed commercial feed whilst grazing on pasture.
In addition on account of their age the beef cattle would have been
housed indoors the previous winter so had greater exposure to commercial feed relatively low in PCDD/Fs and PCBs than the younger
lambs.
The milk data is notable as clear contrasts between ood-prone vs
control sites are evident in 19981999 but not in 20082010. This
may be because levels of PCDD/F and PCBs in 20082010 are a half to
a third the levels observed in 19981999 in milk. This decline was supported by two farms sampled in both time periods. Other studies have
indicated similar declines in retail milk over the same time period
(e.g. France; Durand et al., 2008; UK; Food Standards Agency, 2003;
Food Standards Agency, 2007). During the 1990s, tighter controls on industrial facilities, led to around 7080% reductions in atmospheric emissions of PCDD/Fs and PCBs across the EU (Dore et al., 2003). Since 2000,
the period covered by this study, emissions have reduced more slowly

114

I.R. Lake et al. / Environment International 77 (2015) 106115

(DEFRA, 2012). Hence a reduction in environmental contamination between 19981999 and 20082010 is unlikely to have played a major
role in the declines observed in our milk data. This is supported by our
environmental data, and there were no signicant differences between
soil and grass levels between 19981999 and 20082010. Previous research has pointed to a general decline in PCDD/F and PCBs in soils
and grass levels in the UK (Environment Agency, 2007) but this study
was comparing an earlier point in time (early 1990s) before emissions
stabilised.
One reason for these reductions in milk may be the lower levels of
PCDD/Fs and PCBs in commercial feed. Between 19981999 and 2008
2010 new regulations resulted in tighter controls of PCDD/Fs and PCBs
in animal feed (EC, 2006). However, the small reduction in summed
TEQ in commercial feed (0.070.06 ng TEQ/kg fat) suggests that other factors may be important. Changes to dairy cattle animal husbandry may explain these differences. Between 19981999 and 20082010 the number
of dairy farms in the UK halved, the average size of dairy herds increased
(DairyCo., 2013b) and the average milk production of a dairy cow increased by nearly 18% (DairyCo., 2013a). Larger herds are usually managed more intensively, and consequently utilise more commercial feed
to produce more milk per animal (Promar International, 2010). Over
this period the consumption of commercial feed in dairy cattle increased
over 43% (Promar International, 2010). Data from this study indicates that
commercial feed has lower levels of PCDD/Fs and PCBs in comparison to
grass (summed TEQ; River Trent commercial feed 0.07 ng TEQ/kg dry [assuming approximately 12% moisture] vs. 0.27 ng TEQ/kg dry in grass). In
additional commercial feed is more nutrient dense and less likely to be
contaminated with soil. These factors in combination indicate that increases in the consumption of commercial feed are a plausible explanation for the decline in PCDD/Fs and PCBs observed in milk between
19981999 and 20082010.
5. Conclusions
This paper has demonstrated that regular river ooding events
transfer PCDD/Fs and PCBs to the environment (soil and grass) of industrial river catchments. It has also indicated that such contamination can
be transferred to food. For context we have previously reported that in
terms of levels found in food some of the 20082010 beef samples
were above the action levels or maximum levels under EU regulations
applicable at the time of sampling (Lake et al., 2014) and some of the
milk samples from 19981999 would also have fallen into this category.
The lamb and milk samples from 20082010 were all below the action
levels or maximum levels under EU regulations applicable at the time of
sampling. Although ooding can be a mechanism transferring PCDD/Fs
and PCBs to food, we have shown that the impact varies by food type
(e.g. an effect was seen for beef but not lamb. For milk an effect was
seen in the late 1990s but not more recently). This may be due to physiological differences between animals, the age of the animals, as well as
differences in animal husbandry, specically the amount of commercial
feed fed to animals. Finally it suggests that although environmental contamination (levels in soil and grass associated with river sediment) has
not changed in the ten years since the late 1990s the levels of PCDD/Fs
and PCBs in milk have reduced. We suggest that this is due to the increasing use of commercial feed in milk production. These results have
several implications for the risk management of food produced on
ood-prone land in catchments with a history of industrialisation. The
rst is that the choice of farming may have a large impact on levels of
contaminants in food. Using data from three production systems on
the Trent in 20082010, we indicate that PCDD/Fs appear to be accumulated to a greater degree in beef than in lamb, and the production of milk
leads to lower levels of both PCDD/Fs and PCBs in food compared to beef
production. However, the milk data illustrates our second point, the
importance of changing animal husbandry practices. We argue that
the increasing amounts of commercial feed given to dairy cattle to
raise milk yields, together with regulations lowering concentrations of

PCDD/Fs and PCBs in commercial animal feed, appear to be one factor

behind the reduced levels of PCDD/Fs and PCBs in milk. This is observed
over a period when environmental levels (soil and grass) have remained
relatively constant. Such approaches may be used to manage the risks of
food produced on all contaminated land (including ood-prone contaminated land) worldwide. Results also suggest that the exclusion of livestock from ood-prone areas of industrial river catchments may be an
important risk management strategy.
Acknowledgements
We gratefully acknowledge the United Kingdom Food Standards
Agency (CO 1044 and CO1037) for funding these studies. We would
also like to thank Milk Marque for helping to arrange initial contacts
with dairy farmers and to all the farmers who permitted us to take samples of milk, beef, lamb, soil and grass. Dr Lake is part funded by The National Institute for Health Research, Health Protection Research Unit in
Emergency Preparedness and Response at King's College London. The
opinions and conclusions expressed in this article are solely the views
of the author(s) and do not necessarily reect those of the Food Standards Agency.
Appendix A. Supplementary data
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.envint.2015.01.006.
References
Alcock, R.E., Jones, K.C., 1996. Dioxins in the environment: a review of trend data. Environ.
Sci. Technol. 30, 31333143.
Alivernini, S., Battistelli, C.L., Turrio-Baldassarri, L., 2011. Human milk as a vector and an
indicator of exposure to PCBs and PBDEs: temporal trend of samples collected in
Rome. Bull. Environ. Contam. Toxicol. 87, 2125.
Boxall, A.B.A., Hardy, A., Beulke, S., Boucard, T., Burgin, L., Falloon, P.D., Haygarth, P.M.,
Hutchinson, T., Kovats, R.S., Leonardi, G., Levy, L.S., Nichols, G., Parsons, S.A., Potts,
L., Stone, D., Topp, E., Turley, D.B., Walsh, K., Wellington, E.M.H., Williams, R.J., 2009.
Impacts of climate change on indirect human exposure to pathogens and chemicals
from agriculture. Environ. Health Perspect. 17, 508514.
DairyCo., 2013a. Average Milk Yield.
DairyCo., 2013b. The Structure of the GB Dairy Farming Industry What Drives Change?
University of Nottingham and The Andersons Centre
DEFRA, 2002. Dioxins and Dioxin-like PCBs in the UK environment: Consultation Document.
Department for Environment, Food and Rural Affairs, London.
DEFRA, 2012. Statistical Release: 14 February 2012 Emissions of Air Pollutants in the UK,
1970 to 2010.
Demond, A., Adriaens, P., Towey, T., Chang, S.C., Hong, B., Chen, Q., Chang, C.W., Franzblau,
A., Garabrant, D., Gillespie, B., Hedgeman, E., Knutson, K., Lee, C.Y., Lepkowski, J.,
Olson, K., Ward, B., Zwica, L., Luksemburg, W., Maier, M., 2008. Statistical comparison
of residential soil concentrations of PCDDs, PCDFs, and PCBs from two communities
in Michigan. Environ. Sci. Technol. 42, 54415448.
Dore, C.J., Goodwin, J.W.L., Watterson, J.D., Murrells, T.P., Passant, N.R., Hobson, M.M.,
Haigh, K.E., Baggott, S.L., Pye, S.T., Coleman, P.J., King, K.R., 2003. UK Emissions of
Air Pollutants 1970 to 2001. DEFRA, London.
Durand, B., Dufour, B., Fraisse, D., Defour, S., Duhem, K., Le-Barillec, K., 2008. Levels of
PCDDs, PCDFs and dioxin-like PCBs in raw cow's milk collected in France in 2006.
Chemosphere 70, 689693.
EC, 2006. Commission Recommendation (2006/13/EC) of 4 February 2006 on undesirable
substances in animal feed as regards dioxins and dioxin-like PCBs. Off. J. Eur. Communities L22/44, 514.
Environment Agency, 2007. UK Soil and Herbage Pollutant Survey; UKSHS Report No. 10
Environmental Concentrations of Polychlorinated Dibenzo-p-dioxins and
Polychlorinated Dibenzofurans in UK Soil and Herbage. Environment Agency, Bristol.
EU, 2002. Directive 2002/32/EC of the European Parliament and of the Council on Undesirable Substances in Animal Feed.
EU, 2006. Commission Directive 2006/13/EC of 3 February 2006 Amending Annexes I and
II to Directive 2002/32/EC of the European Parliament and of the Council on Undesirable
Substances in Animal Feed as Regards Dioxins and Dioxin-like PCBs.
EU, 2012. Commission Regulation (EU) No 277/2012 of 28 March 2012 Amending Annexes I and II to Directive 2002/32/EC of the European Parliament and of the Council
as Regards Maximum Levels and Action Thresholds for Dioxins and Polychlorinated
Biphenyls.
Fattore, E., Vigano, L., Mariani, G., Guzzi, A., Benfenati, E., Fanelli, R., 2002. Polychlorinated
dibenzo-p-dioxins and dibenzofurans in River Po sediments. Chemosphere 49,
749754.
Fernandes, A., White, S., D'Silva, K., Rose, M., 2004. Simultaneous determination of PCDDs,
PCDFs, PCBs and PBDEs in food. Talanta 63, 11471155.

I.R. Lake et al. / Environment International 77 (2015) 106115

Food Standards Agency, 2003. Dioxins and Dioxin-like PCBs in the UK Diet: 2001 Total
Diet Study Samples. UK Food Standards Agency, London.
Food Standards Agency, 2007. Food Standards Agency. Dioxins and Dioxin-like PCBs in
Foods EU Monitoring 2006 Food Survey Information Sheet FSES04/07.
Foxall, C.D., Fernandes, A., Lake, I.R., White, S., Lovett, A.A., Rose, M., 2004. The Effects of
River Flooding on Dioxin and PCB Levels in Pastureland Soil, Grass and Cow's Milk.
Food Standards Agency, London.
Harrison, N., Wearne, S., Gem, M.G.D.M., Gleadle, A., Startin, J., Thorpe, S., Wright, C., Kelly, M.,
Robinson, C., White, S., Hardy, D., Edinburgh, V., 1998. Time trends in human dietary
exposure to PCDDs, PCDFs and PCBs in the UK. Chemosphere 37, 16571670.
Healey, W.B., 1972. Ingested soil and animal nutrition. Proceedings of the New Zealand
Grassland Association 34, pp. 8084.
Hedgeman, E., Chen, Q., Hong, B., Chang, C.W., Olson, K., LaDronka, K., Ward, B., Adriaens,
P., Demond, A., Gillespie, B.W., Lepkowski, J., Franzblau, A., Garabrant, D.H., 2009. The
University of Michigan dioxin exposure study: population survey results and serum
concentrations for polychlorinated dioxins, furans, and biphenyls. Environ. Health
Perspect. 117, 811817.
Hendriks, A.J., Wever, H., Olie, K., Van De Guchte, K., Liem, A.K.D., Van Oosterom, R.A.A.,
Van Zorge, J., 1996. Monitoring and estimating concentrations of polychlorinated
biphenyls, dioxins, and furans in cattle milk and soils of rhine-delta oodplains.
Arch. Environ. Contam. Toxicol. 31, 263270.
Hilscherova, K., Dusek, L., Kubik, V., Cupr, P., Hofman, J., Klanova, J., Holoubek, I., 2007.
Redistribution of organic pollutants in river sediments and alluvial soils related to
major oods. J. Soils Sediment. 7, 167177.
Lake, I.R., Foxall, C.D., Lovett, A.A., Fernandes, A., Dowding, A., White, S., Rose, M., 2005.
Effects of river ooding on PCDD/F and PCB levels in cows' milk, soil, and grass. Environ.
Sci. Technol. 39, 90339038.
Lake, I.R., Fernandes, A., Foxall, C.D., Rose, M., Lewis, M., White, O., 2012. The Impact of
River Flooding on Food Produced on Flood Plains. Food Standards Agency, London.
Lake, I.R., Fernandes, A., Foxall, C.D., Rose, M., Lewis, M., White, O., Dowding, A., 2014. The
effects of river ooding on PCDD/F and PCB levels in beef. Sci. Total Environ. 491-492,
184191.
Liem, A.K.D., Frst, P., Rappe, C., 2000. Exposure of populations to dioxins and related
compounds. Food Addit. Contam. 17, 241259.
Lignell, S., Aune, M., Darnerud, P.O., Cnattingius, S., Glynn, A., 2009. Persistent organochlorine
and organobromine compounds in mother's milk from Sweden 19962006:
compound-specic temporal trends. Environ. Res. 109, 760767.

115

Lorber, M., Saunders, P., Ferrario, J., Leese, W., Winters, D., Cleverly, D., Schaum, J., Deyrup,
C., Ellis, R., Walcott, J., Dupuy, A., Byrne, C., McDaniel, D., 1997. A statistical survey of
dioxin-like compounds in United States pork fat. Presented at Dioxin '97. 17th International Symposium on Chlorinated Dioxins and Related Compounds, Held August
2529 at Indianapolis, USA. Organohalogen Compd., p. 32
Norwegian Institute of Public Health, 2009. Interlaboratory Comparison on Dioxins in
Food Sixth Round of an International Study. Norwegian Institute of Public Health,
Oslo, Norway.
Promar International, 2010. Milkminder Newsletter January 2010. p. 142.
Schecter, A., Gasiewicz, T.A. (Eds.), 2003. Dioxins and Health, 2nd edition John Wiley &
Sons, London and New York.
Schuster, J.K., Gioia, R., Moeckel, C., Agarwal, T., Bucheli, T.D., Breivik, K., Steinnes, E., Jones,
K.C., 2011. Has the burden and distribution of PCBs and PBDEs changed in European
background soils between 1998 and 2008? Implications for sources and processes.
Environ. Sci. Technol. 45, 72917297.
Shen, H., Han, J., Tie, X., Xu, W., Ren, Y., Ye, C., 2009. Polychlorinated dibenzo-p-dioxins/
furans and polychlorinated biphenyls in human adipose tissue from Zhejiang Province,
China. Chemosphere 74, 384388.
Sun, S.J., Kayama, F., Zhao, J.H., Ge, J., Yang, Y.X., Fukatsu, H., Iida, T., Terada, M., Liu, D.W.,
2011. Longitudinal increases in PCDD/F and dl-PCB concentrations in human milk in
northern China. Chemosphere 85, 448453.
van den Berg, M., Birnbaum, L., Bosveld, A.T.C., Brunstrm, B., Cook, P., Feeley, M., Giesy,
J.P., Hanberg, A., Hasegawa, R., Kennedy, S.W., Kubiak, T., Larsen, J.C., Van Leeuwen,
F.X.R., Liem, A.K.D., Nolt, C., Peterson, R.E., Poellinger, L., Safe, S., Schrenk, D., Tillitt,
D., Tysklind, M., Younes, M., Wrn, F., Zacharewski, T., 1998. Toxic equivalency
factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife. Environ. Health
Perspect. 106, 775792.
Xia, C., Lam, J.C.W., Wu, X., Xie, Z., Lam, P.K.S., 2012. Polychlorinated biphenyls (PCBs) in
marine shes from China: levels, distribution and risk assessment. Chemosphere 89,
944949.
Zhang, H., Ni, Y., Chen, J., Su, F., Lu, X., Zhao, L., Zhang, Q., Zhang, X., 2008. Polychlorinated
dibenzo-p-dioxins and dibenzofurans in soils and sediments from Daliao River Basin,
China. Chemosphere 73, 16401648.
Zhao, B., Zheng, M., Jiang, G., 2011. Dioxin emissions and human exposure in China:
a brief history of policy and research. Environ. Health Perspect. 119, A112A113.