Not to be quoted without prior reference to the authors
Fisheries Research Services Contract Report No 05/04
BACKGROUND/REFERENCE CONCENTRATIONS (BRCS)
FOR THE UK
Ian M Davies
March 2004
Fisheries Research Services
Marine Laboratory
Victoria Road
Aberdeen AB11 9DB
BACKGROUND/REFERENCE CONCENTRATIONS
(BRCS) FOR THE UK
Ian M Davies
Fisheries Research Services, Marine Laboratory
375 Victoria Road, Aberdeen, AB11 9DB
1. BACKGROUND
Chemical data on hazardous substances submitted to OSPAR programmes and used in
OSPAR Status Reports, etc are assessed against Background/Reference Concentrations
(BRCs) and Ecotoxicological Assessment Criteria (EACs). The UK has generally expressed
reservations over aspects of this process, emphasising the difficulties in deriving and
applying such assessment criteria over wide geographical areas. In particular, BRC values
need to take account of natural variability, for example, in the source rocks for coastal
sediments. The UK has suggested that assessments made against the current OSPAR
BRCs could therefore be misleading.
OSPAR has recognised that the current values require revision, and MON 2002 put in place
a process that will lead to a Workshop in this area during 2004. The definition of BRC
values appropriate to the UK will therefore strengthen the UK position at this Workshop and
in subsequent discussions.
The aim of this project is to propose Background/Reference Concentrations for OSPAR
priority contaminants the relevant environmental matrices.
2. INTRODUCTION
Background/reference concentrations (BRC) and ecotoxicological assessment criteria (EAC)
have been central to the OSPAR process of data evaluation. Comparison with these
concentrations is the starting point for the discussion of chemical monitoring data in terms of
whether the concentrations found in the environment represent a degree of contamination,
and whether pollution (in the sense of biological harm) is likely to be occurring.
EACs are derived from toxicological information. Comparisons of field data with EACs are
designed to provide information on whether particular contaminants are present in the
environment at concentrations high enough to potentially lead to deleterious biological
effects.
In contrast, comparisons with BRC values indicate whether contamination can be detected,
but do not provide any information on the biological significance of that contamination.
OSPAR originally adopted assessment criteria for its international monitoring programmes
through a desire to express significance of chemical monitoring results and to compare
monitoring data between areas and times. The underlying objectives of these processes
within the OSPAR community was to recognise the presence (and thereby the responsibility)
for pollution, to identify the need control measures, and subsequently to assess the
effectiveness of any control measures that were adopted.
1
Background/Reference Concentrations (BRCs) For The UK
As indicated above, the primary assessment tools currently used by OSPAR are EACs and
BRCs. Their simplest method of application is to derive statements as to whether observed
concentrations in the environment are above or below the appropriate values. Another
important element of OSPAR monitoring activity is the measurement of temporal trends in
concentrations of contaminants or degrees of biological effects. As well as leading to
statements that concentrations are increasing or decreasing, it is possible to combine
temporal trend information with EAC and BRC values to indicate the potential significance of
any observed trends.
Ecotoxicological Assessment Criteria (EAC) are used to derive expressions of
biological/ecotoxicological risk. They are described by OSPAR as “concentrations below
which no harm is expected”, and “above which concern is indicated”.
Background/Reference Concentrations (BRC) are intended to provide baseline or reference
concentrations, and to describe environmental conditions under which there are no
anthropogenic influence on the concentrations of the target substances in the environment.
The current OSPAR BRC values were mainly developed at a Workshop held in Hamburg in
1996. The Workshop discussed at some length the possible approaches to the derivation of
appropriate BRCs. The main difficulties are that most environmental materials available for
analysis today, such as water, fish or shellfish, in most parts of the OSPAR area will reflect
current environmental conditions, together with the current levels of environmental
contamination. The presence of persistent organic contaminants in remote (e.g. Arctic)
areas is clear evidence of long-range transport of these substances away from areas where
they are manufactured and used, and it is known that other substances, such as lead and
cadmium, can exhibit significant atmospheric transport. The Workshop concluded that there
were three main approaches that could be adopted:
a)
Background concentrations could be derived from data from geological times, i.e.
pre-industrial times. Clearly, there is no opportunity to collect material today from
most environmental matrices that would reflect pre-industrial conditions. The one
possible opportunity are sub-surface sediments preserved in areas where sediment
accumulation is occurring. This approach has been demonstrated to be effective in
specialised areas such as Norwegian fjords, but there are rather few areas in UK
marine waters where continuous accumulation of sediment has occurred over the
necessary time scale. Areas which could probably be used in this way are some
Scottish sea lochs, and a few off-shore accumulating areas, such as basins on the
west coat of Scotland and mud banks in the Irish Sea. There are few, if any, such
locations off the east coast of Great Britain.
b)
Background concentrations could be derived from published historical data, i.e.
historical measurements.
The main problems with this approach are the
unavailability of any data for many of the OSPAR priority substances (such as metals
and persistent organic contaminants) more than a few decades old, and concerns
over the quality of the data that may be available. The only substances for which
useful and reliable data may be available are nutrients in sea water, but these will not
extend back beyond the early 20th century.
c)
Background concentrations could be derived from recent data from pristine areas, i.e.
areas that do not experience significant anthropogenic contamination. As noted
above, the long-range atmospheric transport of contaminants is well recognised, and
results in the presence of contaminants in areas very distant from sources. Some
environmental processes can lead to the concentration of volatile organic
contaminants in polar areas, areas that might intuitively be considered relatively
2
Background/Reference Concentrations (BRCs) For The UK
unlikely to experience anthropogenic contamination. The widespread distribution of
contaminants leads to the concept of there being a background degree of general
environmental contamination. The level is increased in areas that receive more
direct inputs of contaminants.
In relation to local, national or regional control measures, the general background level of
contamination is much less susceptible to management action than are the contaminant
concentrations in smaller, more localised, areas of contamination. In the latter case, sources
may be identifiable and control measures can be targeted appropriately. In the former case,
the sources of the contaminants may be globally distributed and very diverse.
The procedures that have been adopted for this project therefore mainly rely on approach c)
above, i.e. the identification of ranges of concentrations to reflect the ranges found in
relatively uncontaminated parts of the UK. The UK is fortunate in this respect, in that it
includes considerable areas of sea that do not receive significant direct or riverine
discharges of OSPAR priority contaminants. For example, there are large areas in the north
and far west of the UK where industrial activity and population numbers are low. These
areas, and others, provide opportunities to observe concentrations of contaminants that can
be considered as background values.
3. DERIVATION OF BACKGROUND/REFERENCE CONCENTRATIONS
In this section of the report, combinations of environmental matrices (shellfish tissue, fish
tissue and sediment) and contaminants (metals, CBs and PAH) are considered
systematically and proposed BRC values are developed. The data utilised were provided by
FRS, CEFAS, SEPA and DARD laboratories. All these laboratories contribute to the UK
National Marine Monitoring Programme, and all participate in the QUASIMEME Laboratory
Performance Study scheme.
The expression of the BRC values follows the pattern currently used by OSPAR, i.e. they are
presented as ranges of concentrations (or of ratios of concentrations), and seek to reflect the
variability found naturally in the marine environment. Through experience of using these
values in OSPAR and other assessment exercises, it has become apparent that assessors
wish to make comparisons between concentrations in field samples and the BRC (or EAC)
values. Expression of the BRC as a range is not easily accommodated in statistical tests for
significance of differences. It is likely that the details of how these comparisons might be
made will be discussed at the OSPAR BRC Workshop planned for February 2004. It may be
that a different formulation of BRC values, which is more readily used in statistical tests, will
emerge from these meetings.
3.1
Contaminants in Mussels
3.1.1
Metals
Three sets of data were available for metals in mussel tissue. These were:
•
FRS data for Scotland obtained in 1999–2002 as part of a project in support of
requirements under the Shellfish Hygiene Directive.
•
CEFAS data for England and Wales obtained during 1996–1997, also as part of a
project in support of requirements under the Shellfish Hygiene Directive.
3
Background/Reference Concentrations (BRCs) For The UK
•
SEPA data for Scotland obtained in 1990–2001 during monitoring to meet
requirements under the Shellfish Growing Waters Directive.
The FRS and CEFAS data were on a wet weight basis, those for SEPA on a dry weight
basis. The SEPA data were adjusted to a wet weight basis using a correction factor of 6.
As the data sets were of different sizes, they were treated separately during data analysis.
The FRS data were initially sorted to eliminate data from sites likely to be close to potential
sources of contamination, such as might be found in major estuaries. The remaining data
were therefore from areas of Scotland remote from recognised sources of contamination. As
these locations might all be considered as potentially reflecting reference conditions, the
data were expressed as ranges of values.
Both the CEFAS and SEPA data covered a range of environmental settings, covering
potential reference conditions to areas where some degree of contamination was likely.
These data were not sorted, but an expression of the lower range of values was obtained by
computing the minimum and 30-percentile values of each data set. As it was likely that both
the SEPA and CEFAS data would include more than 30% of sampling points that were
distant from recognised sources of contamination, it was expected that this would yield
relatively low values, possibly falling below the upper limit of an appropriate reference
concentration range.
The summarised data are listed in Table 1. Not all elements were covered in all three
surveys. However, in general, there is quite good agreement between the three sets of data.
As expected, the concentration ranges are similar in all three surveys for elements that are
known to be regulated by mussels, such as copper and zinc. On the other hand, the
concentrations of lead (which is much less well regulated by mussels) reported for England
and Wales are rather greater than those for Scotland, possibly reflecting the impacts of both
waste discharges and mineralisation on coastal waters.
Table 1 also includes proposed BRC values for metals in mussel tissue, derived from the
summarised data sets, and a comparison with the current OSPAR BRC concentrations. For
elements where three sets of data are available, the upper bound to the proposed range is
the median of the upper bound of the individual ranges. In cases where only two sets of
data are available, the upper bound to the proposed range is the higher upper bound of the
two individual ranges.
4
Background/Reference Concentrations (BRCs) For The UK
TABLE 1
Metals in mussel tissue – Ranges of background concentrations derived from FRS, CEFAS
and SEPA data, proposed UK background concentrations and comparisons with the current
OSPAR BRC values.
Element
FRS
mg kg-1
wet weight
CEFAS
mg kg-1
wet weight
As
1 – 3.5
0.9 – 1.9
Cd
0.05 – 0.15
<0.01 – 0.15
Cu
0.8 – 1.5
0.6 – 1.2
Hg
< 0.04
0.01 – 0.02
Pb
0.03 – 0.2
Zn
SEPA
mg kg-1
wet weight
Insufficient
data
0.07 – 0.15
Proposed UK
BRC value
mg kg-1
wet weight
<0.01 – 0.15
Current OSPAR
BRC
mg kg-1
wet weight
No OSPAR BRC
defined
0.07 – 0.11
1 – 3.5
0.6 – 1.2
0.76 – 1.10
0.01 - 0.04
0.005 – 0.010
0.47 – 0.64
0.54 -0.90
Insufficient
data
0.03 – 0.19
0.03 - 0.2
0.010 – 0.19
8 – 15
6.5 – 14
7.8 - 12
6 - 14
Cr
<0.05 – 0.2
0.29 – 0.41
0.06 – 0.29
<0.05 – 0.3
Ag
Insufficient
data
<0.01 –
0.013
Se
0.2 – 0.5
0.38 – 0.61
Insufficient
data
Insufficient
data
Ni
Insufficient
data
0.13 – 0.37
11.6 - 30
No OSPAR BRC
defined
No OSPAR BRC
defined
No OSPAR BRC
defined
No OSPAR BRC
defined
3.1.2
0.03 – 0.15
< 0.01 – 0.13
0.2 - 0.6
0.1 – 0.4
Polycyclic aromatic hydrocarbons (PAH)
Two sets of data were available for polycyclic aromatic hydrocarbons (PAH) in mussel
tissue. These were:
•
FRS data for Scotland obtained in 1999–2002 as part of a project in support of
requirements under the Shellfish Hygiene Directive.
•
CEFAS data for England and Wales obtained during 1996–1997, also as part of a
project in support of requirements under the Shellfish Hygiene Directive.
Both sets of data were on a wet weight basis. As the data sets were of different sizes, they
were treated separately during data analysis. The FRS data were initially sorted to eliminate
data from sites likely to be close to potential sources of contamination, such as might be
found in major estuaries. The remaining data were therefore from areas of Scotland remote
from recognised sources of contamination. As these locations might all be considered as
potentially reflecting reference conditions, the data were expressed as ranges of values.
The CEFAS data covered a range of environmental settings, including potential reference
conditions and areas where some degree of contamination was likely. These data were not
sorted, but an expression of the lower range of values was obtained by computing the
minimum and 30-percentile values for each determinand.
The summarised data for the OSPAR priority PAH compounds are listed in Table 2. In
general, there is quite good agreement between the two sets of data, bearing in mind the
5
Background/Reference Concentrations (BRCs) For The UK
wide ranges of concentrations of PAH compounds that can be found in mussels. For some
compounds, the upper bound for the FRS data exceeds that for the CEFAS data, and for
other compounds the reverse occurs.
Table 2 also includes proposed BRC values for PAHs in mussel tissue, derived from the
summarised data sets. In cases where two sets of data are available, the upper bound to
the proposed range is the higher upper bound of the two individual ranges.
There are no current OSPAR BRCs for PAH in mussels.
TABLE 2
Polycyclic aromatic hydrocarbons in mussel tissue – Ranges of background concentrations,
derived from FRS and CEFAS data, and proposed UK background concentrations. There
are no current OSPAR BRCs for PAH in mussels.
FRS
ng g-1
wet weight
CEFAS
ng g-1
wet weight
Proposed
UK BRC
values
ng g-1
wet weight
Current OSPAR BRC
ng g-1
wet weight
Naphthalene
0.2 – 1.6
0.2 – 1.6
0.2 – 1.6
Not defined
Phenanthrene
0.5 – 2.5
0.6 – 3.5
0.5 – 3.5
Not defined
Anthracene
0.2 – 0.8
0.1 – 0.5
0.2 – 0.8
Not defined
Fluoranthene
0.6 – 5.6
2.0 – 14
0.6 - 14
Not defined
Pyrene
0.7 – 5.1
1.4 – 11
0.7 - 11
Not defined
Benzo[a]anthracene
0.2 – 4.0
0.5 – 2.0
0.2 – 4.0
Not defined
Chrysene
0.4 – 7.3
1.5 – 4.0
0.4 – 7.3
Not defined
Benzo[a]pyrene
0.2 – 2.3
0.2 – 0.9
0.2 – 2.3
Not defined
Benzo[ghi]perylene
0.2 – 5.0
0.1 – 1.3
0.1 – 5.0
Not defined
Indeno[123-cd]pyrene
0.2 – 1.6
0.1 – 0.6
0.1 – 1.6
Not defined
Determinand
3.1.3
Chlorobiphenyls (CB) compounds in mussel tissue
Two sets of data were available for chlorobiphenyl (CB) compounds in mussel tissue. These
were:
•
FRS data for Scotland obtained in 1999–2002 as part of a project in support of
requirements under the Shellfish Hygiene Directive.
•
CEFAS data for England and Wales obtained during 1996–1997, also as part of a
project in support of requirements under the Shellfish Hygiene Directive.
Both sets of data were on a wet weight basis. The reporting limit for the CEFAS data was
I ng/g, and a high proportion of the data were at, or very slightly above, this limit. By
contrast, the FRS used a reporting limit at least 10 times lower, and commonly reported
values between 0.1 and 1 ng/g. The FRS data were therefore considered to be more
6
Background/Reference Concentrations (BRCs) For The UK
reliable at the low concentrations found in mussels from areas distant from recognised
sources of contamination.
As for other determinands, the FRS data were initially sorted to eliminate data from sites
likely to be close to potential sources of contamination, such as might be found in major
estuaries. The remaining data were therefore from areas of Scotland remote from
recognised sources of contamination. As these locations might all be considered as
potentially reflecting reference conditions, the data were expressed as ranges of values.
These values (Table 3) are proposed as BRC for CBs in UK mussels.
TABLE 3
Chlorobiphenyls (CBs) in mussel tissue – Ranges of background concentrations derived
from FRS data, proposed UK background concentrations and comparisons with the current
OSPAR BRC values.
FRS
ng g-1
wet weight
Proposed UK BRC
values
ng g-1 wet weight
Current OSPAR BRC
ng g-1 wet weight
CB28
0.1 – 0.15
0.1 – 0.15
No OSPAR BRC defined
CB52
0.1 – 0.5
0.1 – 0.5
No OSPAR BRC defined
CB101
0.2 – 0.5
0.2 – 0.5
No OSPAR BRC defined
CB118
0.1 – 1
0.1 – 1
No OSPAR BRC defined
CB153
0.1 – 0.3
0.1 – 0.3
0.1 – 0.5
CB138
0.1 – 0.2
0.1 – 0.2
No OSPAR BRC defined
CB180
<0.03 – 0.1
<0.03 – 0.1
No OSPAR BRC defined
0.5 - 2
0.5 - 2
0.35 – 1.7
Determinand
Sum ICES 7 CBs
The same FRS data set included information on the concentrations of selected pesticides,
primarily HCH isomers and compounds in the DDT group. The same analysis process has
been applied to these data to derive possible UK BRC values, and these are tabulated in
Table 4. OSPAR have not set BRC values for pesticides in mussel tissue.
7
Background/Reference Concentrations (BRCs) For The UK
TABLE 4
Selected pesticides in mussel tissue – Ranges of background concentrations derived from
FRS data, proposed UK background concentrations and comparisons with the current
OSPAR BRC values. Blank fields indicate insufficient data are available.
FRS
ng g-1 wet weight
Proposed UK
BRC values
ng g-1
wet weight
Current OSPAR BRC
ng g-1 wet weight
HCH – alpha
0.05 – 0.1
0.05 – 0.1
No OSPAR BRC defined
HCH – beta
No data
Determinand
HCH - gamma
HCH - delta
No OSPAR BRC defined
0.08 – 0.15
0.08 – 0.15
No data
No OSPAR BRC defined
No OSPAR BRC defined
op-DDT
< 0.05 – 0.15
< 0.05 – 0.15
No OSPAR BRC defined
pp-DDT
0.05 – 0.2
0.05 – 0.2
No OSPAR BRC defined
pp-TDE (pp-DDD)
<0.05 – 0.1
<0.05 – 0.1
No OSPAR BRC defined
pp-DDE
0.1 – 0.5
0.1 – 0.5
No OSPAR BRC defined
Dieldrin
<0.05 – 0.5
<0.05 – 0.5
No OSPAR BRC defined
3.1.4
Application of proposed BRC values to NMMP data sets
As part of the process leading to the publication of their latest periodic report, the UK
National Marine Monitoring Programme has compiled and reviewed all the data submitted to
them on contaminants in mussels. The data were filtered for quality, according to the
scheme developed by the Chemical Analytical Quality Control sub-group of the NMMP. The
NMMP data set covered samples from England, Scotland, Wales and Northern Ireland.
As a preliminary exploration of the possible implications of the proposed BRC values for
contaminants in mussel tissue, the final filtered data have been compared to the proposed
BRC values. Tabulations below show the percentages of the filtered data which fall below
the proposed BRC range, within that range, and above that range for metals, CBs and PAHs
in mussel tissue.
8
Background/Reference Concentrations (BRCs) For The UK
TABLE 5
Application of proposed BRC values to filtered NMMP data on metals in mussel tissue
Element
Number of
data points
Percentage of
data points below
BRC range
Percentage of data
points within BRC
range
Percentage of data
points above BRC
range
As
174
3.4
89.7
6.9
Cd
193
0
20.7
79.3
Cu
193
0.5
35.3
64.2
Hg
205
0.5
61.9
37.6
Pb
193
0
4.7
95.3
Zn
205
0.5
25.8
73.7
Cr
205
0
24.9
75.1
Ag
94
1.0
73.5
25.5
Ni
193
0
58.8
41.2
TABLE 6
Application of proposed BRC values to filtered NMMP data on CBs in mussel tissue
CB congener
Number of
data
points
Percentage
of data
points below
BRC range
Percentage of data
points within BRC
range
Percentage of data
points above BRC
range
CB28
164
1.2
29.3
69.4
CB52
171
1.2
58.4
40.4
CB101
182
8.8
18.1
73.1
CB118
174
0
61.5
38.5
CB153
171
0
3.5
96.5
CB138
182
0
1.1
98.9
CB180
183
0
1.1
98.9
9
Background/Reference Concentrations (BRCs) For The UK
TABLE 7
Application of proposed BRC values to filtered NMMP data on PAHs in mussel tissue.
Number
of data
points
Percentage of
data points
below BRC
range
Percentage of
data points
within BRC
range
Percentage of data
points above BRC
range
Naphthalene
13
0
53.8
46.2
Phenanthrene
101
0
22.8
77.2
Anthracene
37
0
81.1
18.9
Fluoranthene
102
0
67.6
32.4
Pyrene
102
1.0
34.3
64.7
Benz[a]anthracene
38
0
84.2
15.8
Chrysene
102
0
45.1
54.9
Benzo[a]pyrene
38
0
71.1
28.9
Benzo[ghi]perylene
101
0
65.3
34.7
Indenopyrene
34
0
97.1
2.9
PAH
A further data set for PAH in mussels was similarly explored. This data set consisted of
analyses of 44 samples of farmed mussels collected at approximately monthly intervals from
Loch Etive, Scotland. Loch Etive is distant from any industrial or urban inputs of PAH, and
experiences only minor boat traffic. The data are therefore expected to represent
background conditions, with the only likely sources of hydrocarbon contamination being fuel
used on the small boat servicing the shellfish farm.
For almost all compounds, the large majority of the data points fall within the proposed BRC
ranges. The points falling below the range were normally samples for which the
concentration was below the limit of quantification of the analytical method. The occurrence
of a moderate proportion of concentrations above the proposed BRC range for naphthalene
would be consistent with occasional low levels of contamination of boat engine fuel.
10
Background/Reference Concentrations (BRCs) For The UK
TABLE 8
Application of proposed BRC values to data on PAHs in mussel tissue from Loch Etive,
Scotland
Number
of data
points
Percentage of
data points
below BRC
range
Percentage of
data points within
BRC range
Percentage of data
points above BRC
range
Naphthalene
44
0
73
27
Phenanthrene
44
2
98
0
Anthracene
44
66
32
2
Fluoranthene
44
2
98
0
Pyrene
44
9
91
0
Benz[a]anthracene
44
18
80
2
Chrysene
44
11
87
2
Benzo[a]pyrene
44
27
71
2
Benzo[ghi]perylene
44
2
96
2
Indenopyrene
44
11
82
7
PAH
3.2
Contaminants in Fish Tissues
3.2.1
Metals in fish muscle
A single set of data has been used for the concentrations of metals (mercury and arsenic) in
fish tissue. These were FRS data for Scotland obtained in relation to the NMMP
programme. The FRS data were initially sorted to eliminate data from sites likely to be close
to potential sources of contamination, such as might be found in major estuaries. The
remaining data were therefore from areas of Scotland remote from recognised sources of
contamination. As these locations might all be considered as potentially reflecting reference
conditions, the data were expressed as ranges of values.
The summarised data are listed in Table 9. Table 9 also includes proposed BRC values for
metals in fish muscle tissue, derived from the summarised data set.
The concentrations of both mercury and arsenic in fish (plaice) muscle are not ideal as
expressions of background concentrations. Firstly, it has been known for many years that
mercury concentrations generally increase with size and age of fish. Concentrations of
mercury in plaice are generally well below any dietary limits, but still cover a considerable
range of values. The expression of a BRC as 0-0.1 mg/kg probably does not provide high
discriminatory power. It might be possible improve the power (reduce the range) by strict
definition of the age/size of fish to be analysed.
Plaice, and other flatfish, are known to exhibit high concentrations of arsenic. This is thought
to be derived from dietary sources, and to accumulate by natural processes. Particularly
high concentrations of arsenic have been reported in flatfish from the central/northern North
Sea, where no significant anthropogenic sources of arsenic are recognised. It may therefore
be appropriate to consider an alternative species for assessment of arsenic contamination.
11
Background/Reference Concentrations (BRCs) For The UK
TABLE 9
Metals (arsenic and mercury) in fish (plaice) muscle tissue – Ranges of background
concentrations derived from FRS data, proposed UK background concentrations and
comparisons with the current OSPAR BRC values for round fish (cod, hake) and flat fish
(dab, flounder, plaice).
3.2.2
Element
Concentration
mg kg-1
wet weight
Proposed UK BRC
values mg kg-1
wet weight
As
3 - 30
3 - 30
Hg
0 - 0.1
0 - 0.1
Current OSPAR BRC
ng g-1 wet weight
0.01 – 0.05 (round fish)
0.03 – 0.07 (flat fish)
Metals in fish liver
Two sets of data were available for metals (lead and cadmium) in fish liver. These were:
•
FRS data for Scotland obtained in relation to the NMMP programme.
•
CEFAS data for England and Wales, also obtained in relation to the NMMP
programme.
The FRS and CEFAS data were both on a wet weight basis.
As the data sets were of different sizes, they were treated separately during data analysis.
The FRS data were initially sorted to eliminate data from sites likely to be close to potential
sources of contamination, such as might be found in major estuaries. The remaining data
were therefore from areas of Scotland remote from recognised sources of contamination. As
these locations might all be considered as potentially reflecting reference conditions, the
data were expressed as ranges of values.
The CEFAS data included a range of environmental settings, covering potential reference
conditions to areas where some degree of contamination was likely. These data were not
sorted, but an expression of the lower range of values was obtained by computing the
minimum and 30-percentile values of each data set. As it was likely that both the SEPA and
CEFAS data would include more than 30% of sampling points that were distant from
recognised sources of contamination, it was expected that this would yield relatively low
values, possibly falling below the upper limit of an appropriate reference concentration
range.
The summarised data are listed in Table 10. The concentration ranges are similar in both
surveys for both lead and cadmium. Table 10 also includes proposed BRC values for metals
in fish liver, derived from the summarised data sets. The upper bound to the proposed
range is the higher upper bound of the two individual ranges, and the lower reflects the
lower, lower bound of the ranges.
12
Background/Reference Concentrations (BRCs) For The UK
TABLE 10
Metals (cadmium and lead) in fish muscle tissue – Ranges of background concentrations
derived from FRS and CEFAS data, and proposed UK background concentrations.
3.2.3
Element
FRS (plaice)
mg kg-1 wet weight
CEFAS (various)
mg kg-1 wet weight
Proposed UK BRC values
mg kg-1 wet weight
Cd
0.04 – 0.08
0.01 – 0.06
0.01 – 0.08
Pb
0.03 – 0.1
0 – 0.1
0 – 0.1
Chlorobiphenyl (CB) compounds in fish liver
Two sets of data were available for chlorobiphenyls (CB) compounds in fish liver. These
were:
•
FRS data for Scotland obtained in relation to the NMMP programme.
•
CEFAS and DARD data for England, Wales and Northern Ireland, also obtained in
relation to the NMMP programme.
The FRS and CEFAS/DARD data were both on a wet weight basis and were also
accompanied by total lipid concentrations. The data were recalculated on a lipid weight
basis to normalise for differences in lipid concentrations between samples.
As the data sets were of different sizes, they were treated separately during data analysis.
The FRS data were initially sorted to eliminate data from sites likely to be close to potential
sources of contamination, such as might be found in major estuaries. The remaining data
were therefore from areas of Scotland remote from recognised sources of contamination. As
these locations might all be considered as potentially reflecting reference conditions, the
data were expressed as ranges of values.
The CEFAS/DARD data included a range of environmental settings, covering potential
reference conditions to areas where some degree of contamination was likely. These data
were not sorted, but an expression of the lower range of values was obtained by computing
the 2-percentile and 30-percentile values of each data set. The 2-percentile level was
selected, rather than the minimum, to minimise the impact of a few, apparently outlying, low
values. As it was likely that the CEFAS data would include more than 30% of sampling
points that were distant from recognised sources of contamination, it was expected that this
would yield relatively low values, possibly falling below the upper limit of an appropriate
reference concentration range.
The summarised data are listed in Table 11. The concentration ranges were generally
similar in both surveys, although some areas of difference were noted, for example the
upper bounds for CB52 and CB101. Table 11 also includes proposed BRC values for
metals in fish liver, derived from the summarised data sets. The lower bound to the
proposed range reflects the lower, lower bound of the two individual ranges. The upper
bound reflects the upper bounds of the two individual ranges where these are similar, and
approximates the mean of the two upper bounds where they appear to differ.
There are no current OSPAR BRCs for CBs in fish liver.
13
Background/Reference Concentrations (BRCs) For The UK
TABLE 11
Chlorobiphenyls (CBs) in fish liver – Ranges of background concentrations derived from
FRS and CEFAS/DARD data, and proposed UK background concentrations. OSPAR have
not defined BRC for CBs in fish liver.
FRS (plaice)
ng g-1
lipid weight
CEFAS/DARD
(various)
ng g-1 lipid weight
Proposed UK BRC values
ng g-1
lipid weight
CB28
6 - 12
1-9
1 - 10
CB52
10 - 43
3 - 14
3 - 30
CB101
12 - 41
3 - 36
3 - 40
CB118
11 - 22
3 - 43
3 - 30
CB153
10 - 25
9 - 95
10 - 60
CB138
9 – 15
6 - 12
6 -15
CB180
5 - 10
3 - 30
3 - 20
122 – 200
97 - 394
100 - 300
Determinand
Sum ICES 7 CBs
3.2.4
Application of proposed BRC values to NMMP data sets
As part of the process leading to the publication of their latest periodic report, the UK
National Marine Monitoring Programme has compiled and reviewed all the data submitted to
them on contaminants in fish tissues. The data were filtered for quality, according to the
scheme developed by the Chemical Analytical Quality Control sub-group of the NMMP. The
NMMP data set included data from England, Scotland, Wales and Northern Ireland.
As a preliminary exploration of the possible implications of the proposed BRC values for
contaminants in fish tissue, the final filtered data have been compared to the proposed BRC
values. Tabulations below show the percentages of the filtered data which fall below the
proposed BRC range, within that range, and above that range for metals and CBs in fish
tissue.
TABLE 12
Application of proposed BRC values to filtered NMMP data on metals in fish muscle
Number of
data points
Percentage of
data points below
BRC range
Percentage of data
points within BRC
range
Percentage of data
points above BRC
range
As
262
31
69
0
Hg
510
0
61
39
Element
14
Background/Reference Concentrations (BRCs) For The UK
TABLE 13
Application of proposed BRC values to filtered NMMP data on metals in fish liver
Number of
data points
Percentage of
data points below
BRC range
Percentage of data
points within BRC
range
Percentage of data
points above BRC
range
Cd
395
<1
38
62
Pb
389
0
33
67
Element
TABLE 14
Application of proposed BRC values to filtered NMMP data on CBs in fish liver
Number of
data points
Percentage of
data points
below BRC
range
Percentage of
data points
within BRC
range
Percentage of data
points above BRC
range
CB28
301
6
30
64
CB52
301
6
38
56
CB101
301
4
30
66
CB118
301
7
19
74
CB153
301
4
14
82
CB138
301
6
1
93
CB180
301
5
24
71
Sum ICES7
301
6
30
64
CB congener
3.3
Contaminants in Sediment
There are two main approaches that can be taken to estimating background concentrations
of contaminants in sediments. The first depends on the analysis of core samples, with a
view to determining concentrations preserved in a historical record. The second involves the
estimation of concentrations in current surface sediments in areas relatively remote from
contamination. In both cases, the underlying geology and mineralogy of the sediment will
influence the observed concentrations.
It was decided not to use core samples as the basis for the current exercise, for a variety of
reasons, including:
a)
Suitable locations for the study of historical core samples are not common round the
UK coast. In marine areas, by far the most extensive suitable locations are in
special, atypical areas, such as the deep basins of sea lochs, where conditions are
relatively quiescent and sediment accumulation is occurring. However, large areas
of the coastal sea bed are not currently accumulating sediment and are unsuitable for
examination of core samples in this way. Comparisons between areas would
therefore be subject to considerable uncertainty.
15
Background/Reference Concentrations (BRCs) For The UK
Studies have been carried out in a limited number of estuarine areas, where
sediment accumulation may have occurred. Again, the geographical coverage is
rather restricted.
b)
There is some uncertainty as to the extent of post-depositional migration of
contaminants in sediment, particularly of trace metals in response to changes in
redox potential, and the authigenic formation of minerals such as pyrite. It is possible
that down-core profiles have been significantly disturbed by such processes.
c)
It can be argued that true background concentrations of synthetic organic
contaminants, such as CBs, should be zero, and that such values should therefore
be present in sediment more than 100 years old, or so. To adopt such a value as
background would inevitably lead to suggestions that few, if any, current sediments
exhibit concentrations of these substances close to background.
There might be some merit in selecting background concentrations that correspond to some
particular historical year. However, it is not clear on what basis such a year should be
selected, particularly as it would be desirable to use the same “reference” year for all
contaminants. Some synthetic contaminants (e.g. CBs) have only been released to the
environment during the past few decades. mIn contrast, the rate of mobilisation of other
substances, such as metals, greatly increased at the onset of the industrial revolution. It has
also been argued that mining activity greatly preceded the industrial revolution in some
areas (e.g. exploitation of UK lead ores in Roman times and Cornish tin mining before that).
Alternatively, the gradual clearance of forests and establishment of more settled forms of
agriculture in pre-Roman times altered the rates of terrestrial erosion, and consequently the
rates of input of contaminants to the sea. It is therefore not clear that the selection of a
“reference” year could be robustly justified.
The current NMMP programme is based around the measurement of contaminant
concentrations in surface sediments, and the temporal trends in such concentrations. The
use of concentrations of contaminants in surface sediments in UK waters, and also distant
from recognised sources of contamination, is therefore consistent with the design of the
monitoring programme.
3.3.1
Chlorobiphenyls (CB) compounds in sediment
Chlorobiphenyls are ubiquitous contaminants of sediment throughout the UK. They are
lipophilic and bind strongly to organic matter in sediment. The primary data set that has
been used to derive proposed BRC ranges was obtained from surface sediments in coastal
(as opposed to estuarine) waters in Scotland. The data were expressed on a dry weight
basis, and were supported by measurements of the organic carbon content of the sediment.
The data have been normalised to organic carbon, to reduce the influence of differences in
particle size distributions (and organic carbon content) between samples.
As for other determinands, the data were initially sorted to eliminate data from sites likely to
be close to potential sources of contamination, such as might be found in major estuaries.
The remaining data were therefore from areas of Scotland remote from recognised sources
of contamination. As these locations might all be considered as potentially reflecting
reference conditions, the data were expressed as ranges of values. These values
(Table 15) are proposed as BRC for CBs in UK sediments. OSPAR have adopted ranges of
background concentrations of selected CB in surface sediments in specific regions of the
Convention area (mainly in the north east of the area), as below. The background
concentrations of contaminants are supported by TOC concentrations, and these have been
combined to give the normalised values in the right hand column of Table 16.
16
Background/Reference Concentrations (BRCs) For The UK
TABLE 15
Ranges of background concentration (in pg g-1 dry weight) of HCB, DDE and selected CB in
surface sediments for the application in specific regions of the Convention Area
Area of
application
TOC
CB-28
CB-52
-1
-1
CB-101
{%}
[pg g ]
[pg g ]
[pg g ]
[pg g ]
[pg g ]
[pg g-1]
Norwegian Sea
Iceland Sea /
Norwegian Sea
South Norway/
Skagerak
0.5
<10
<10
12
25
25
15
0.5
<10
<10
16
20
26
<10
0.6
30
35
65
90
120
60
-1
CB-153
-1
CB-138
-1
CB-180
TABLE 16
Chlorobiphenyls (CBs) in sediment – Ranges of background concentrations derived from
FRS data, proposed UK background concentrations, and current OSPAR background
concentrations (for areas specified in Table 15 above).
Concentration
ng g-1 organic
carbon
Proposed UK
background
concentrations
ng g-1 organic
carbon
Current OSPAR BRC range
ng g-1 organic carbon
CB28
3 – 25
3 – 25
<2 - 5
CB52
10 – 200
10 – 200
<2 – 5
CB101
3 – 150
3 – 150
2.4 – 10.8
CB118
3 – 60
3 – 60
Not defined
CB153
3 – 50
3 – 50
4 – 15
CB138
3 – 60
3 – 60
5 – 20
CB180
3 – 20
3 – 20
<2 - 10
50 - 500
50 - 500
Not defined
Determinand
Sum ICES 7 CBs
3.3.2
Polycyclic aromatic hydrocarbons in sediment
PAH compounds are ubiquitous contaminants of sediment throughout the UK. They are
lipophilic and bind strongly to organic matter in sediment. The primary data set that has
been used to derive proposed BRC ranges was obtained from surface sediments in open
water areas coastal (as opposed to estuarine) in Scotland. The data were expressed on a
dry weight basis, and were supported by measurements of the organic carbon content of the
sediment. The data have been normalised to organic carbon, to reduce the influence of
differences in particle size distributions (and organic carbon content) between samples.
Definition of the background concentrations of PAH presents particular problems. It is our
experience that concentrations can vary greatly from area to area, even where the areas
appear not to be receiving discharges containing PAHs. Recent surveys of PAH in Scottish
17
Background/Reference Concentrations (BRCs) For The UK
inshore waters (west coast sea lochs, Shetland voes, Orkney coastal waters) that would be
considered distant from recognised major sources of PAH have found large systematic
differences in concentrations between sampling areas. Total PAH concentrations in excess
of 5000 ng/g have been found in areas that were expected to be free from contamination.
The underlying reasons for this are not clear, but may be related to the ways in which areas
respond to atmospheric inputs of PAH (for example onto catchment areas), in addition to
local factors such as PAH from road traffic, and marine transport activity.
As for other determinands, the data were therefore initially sorted to eliminate data from sites
likely to be close to potential sources of contamination, such as might be found in estuaries
and inshore areas. These locations might all be considered as potentially reflecting
background/reference conditions. The values in Table 18 are the 0–90th percentiles of the
trimmed coastal (offshore) dataset, and the data are expressed as ranges of values. These
values (Table 18) are proposed as BRC for PAHs in UK sediments. The ranges of values
are large, reflecting the high variability in the observed concentrations, even after
normalisation to organic carbon.
OSPAR have adopted a series of ranges of background concentrations of selected PAHs
(together with supporting TOC concentrations) in surface sediments for application in
selected regions of the Convention area. These data have been combined to give
normalised values for the Northern and Southern North Sea area in the right hand columns
of Table 18. In general, the upper bounds of the OSPAR and proposed UK ranges are
similar, but the lower bounds of the UK ranges are often lower than those adopted by
OSPAR for the North Sea.
TABLE 17
Ranges of background concentrations (in ng g-1 dry weight) of selected PAHs in surface
sediments (in brackets TOC percentages) for the application in selected regions of the
Convention area.
PAH
NAPH
NAPHC1
NAPHC2
NAPHC3
ACY
ACE
FL
PHEN
PHENCl
PHENC2
ANT
FLU
PYR
BaA
CHR
Bb+kF
BaP
DBahA
PER
1123P
BghiP
Barents Sea
min
max
<1 (0.1)
<1 (0.1)
<1 (0.1)
<1 (0.1)
0.3 (0.54)
0.1 (0.54)
0.3 (0.46)
<1 (0.1)
<1 (0.1)
<1 (0.1)
0.2 (0.23)
<1 (0.1)
<1 (0.1)
<1 (0.1)
<1 (0.1)
7.4 (0.23)
<1 (0.1)
0.8 (0.23)
<1 (0.1)
<1 (0.1)
<1 (0.1)
370 (1.8)
2025 (1.2)
2329 (1.2)
982 (1.8)
0.8 (0.93)
0.3 (0.23)
86 (2.4)
434 (1.2)
727 (2.4)
642 (1.6)
0.8 (0.77)
87 (2.4)
77 (2.4)
33 (1.8)
217 (2.3)
34.8 (0.93)
41 (2.4)
3.8 (0.77)
287 (1.2)
75 (2.1)
110 (2.3)
Arctic Ocean to
Iceland Sea
min
max
Northern North Sea/
Skagerrak
min
max
Southern North Sea
min
max
1.7 (0.23)
6.7 (0.75)
7.7 (0.63)
62.2 (2.37)
<0.2 (0.05)
45.5 (1.3)
0.3 (0.28)
0.2 (0.75)
0.4 (0.23)
2.3 (0.23)
0.8 (0.75)
0.5 (1.23)
2.6 (1.23)
16.4 (0.75)
0.8 (0.63)
0.5 (0.63)
1.8 (0.63)
12.9 (0.63)
3.8 (2.23)
5.8 (2.37)
16.1 (2.37)
109.9 (2.37)
<0.2 (0.05)
<0.2 (0.05)
<0.2 (0.05)
0.46 (0.05)
4.1 (1.3)
6.4 (1.3)
15.3 (1.3)
60.1 (1.3)
0.2 (0.23)
1.5 (0.23)
1.7 (0.59)
0.8 (0.59)
1.5 (0.23)
7.4 (0.23)
1.0 (0.87)
0.8 (2.50)
0.8 (0.77)
7.5 (0.77)
6.4 (0.77)
4.0 (0.77)
9.4 (0.77)
29.8 (0.77)
3.8 (0.77)
3.6 (0.77)
1.5 (0.63)
13.8 (0.63)
11.3 (0.63)
7.7 (0.63)
12.8 (0.63)
46.3 (0.63)
8.8 (0.63)
5.6 (0.63)
13.8 (2.37)
159.6 (2.37)
128.4 (2.37)
69.0 (2.37)
91.3 (2.37)
433.8 (2.37)
111.6 (2.37)
26.5 (2.23)
<0.2 (0.05)
0.72 (0.05)
0.57 (0.05)
<0.2 (0.05)
<0.2 (0.05)
1.07 (0.05)
<0.2 (0.05)
<0.2 (0.05)
16.0 (1.3)
97.5 (1.3)
77.8 (1.3)
41.7 (1.3)
47.2 (1.3)
141.6 (1.3)
51.0 (1.3)
8.3 (1.3)
6.2 (0.59)
4.1 (0.23)
22.7 (0.77)
19.2 (0.77)
43.4 (0.63)
30.7 (0.63)
211.6 (2.23)
189.5 (2.37)
<0.2 (0.05)
<0.2 (0.05)
69.6 (1.3)
62.9 (1.3)
18
Background/Reference Concentrations (BRCs) For The UK
TABLE 18
Polycyclic aromatic hydrocarbons (PAHs) in sediment – Ranges of background
concentrations derived from FRS data, proposed UK background concentrations, and
current OSPAR background concentrations for selected regions of the Convention area (see
Table 17 above).
Determinand
Concentrations
derived from FRS
data.
-1
Proposed UK
background
concentrations.
OSPAR
background
concentrations.
Northern North
Sea.
OSPAR background
concentrations.
Southern North Sea.
ng g organic
carbon
ng g-1 organic
carbon
Naphthalene
60 – 3000
60 – 3000
1200 - 2600
<400 - 3500
Phenanthrene
150 - 7000
150 - 7000
2000 - 4600
900 - 4600
0 – 800
0 – 800
240 - 580
<400 - 1200
Fluoranthene
150 – 7000
150 – 7000
2200 – 6700
1400 – 7500
Pyrene
150 – 6000
150 – 6000
1800 – 5400
1100 - 6000
Benzo[a]anthracene
0 – 2500
0 – 2500
1200 - 2900
<400 - 3200
Chrysene
0 – 3000
0 – 3000
2000 - 3900
<400 - 3600
Benzo[a]pyrene
150 – 5000
150 – 5000
1400 – 4700
<400 - 3900
Benzo[ghi]perylene
600 – 9000
600 – 9000
4900 - 8000
<400 - 4800
Indeno[123-cd]pyrene
200 – 8000
200 – 8000
6900 - 9500
<400 – 5400
Total (10 compounds)
2500 - 50000
2500 - 50000
Anthracene
3.3.1
-1
ng g organic
carbon
-1
ng g organic
carbon
Metals in sediment
The question of background concentrations of metals in sediments is closely linked to the
problem of normalisation of metal concentrations in sediments. The current advice from
OSPAR on the normalisation of metal concentrations in sediments is contained in Annex 5 to
the JAMP Guidelines for Monitoring Contaminants in Sediments. This Guidelines document
has not yet been issued by OSPAR in its final form, and therefore is included as Appendix 1
to this report.
The problem of assessment of metal concentrations in sediments arises from two
fundamental factors. Firstly, that metals are natural constituents of marine sediments, and
secondly that the concentrations of metals of both natural and anthropogenic origin are
greater in fine grained sediments than in coarser sediments. The spatial patterns of
distribution of metals in sediment are very strongly influenced by the distribution of fine
grained particles in the sediment.
Debate has continued for many years on the most appropriate procedures to adopt to
remove the influence of particle size on metal distribution patterns. The debates have
covered wide areas from selection of appropriate sampling locations, to the appropriate form
of extraction of the metals prior to analysis, to the appropriate pre-treatment of sediment
(e.g. sieving) prior to analysis, and to the mathematical treatment of field data to remove the
influence of particle size distribution.
19
Background/Reference Concentrations (BRCs) For The UK
Previous OSPAR MON assessments of sediment data have used a simple normalisation
procedure, the calculation of ratios between trace elements and aluminium, a conservative
tracer as surrogate for fine grained material. This approach is consistent with the currently
defined OSPAR BRCs for metals in sediment. However, in practice the approach was not
fully satisfactory. The main area of difficulty was in sediments of low aluminium content
(< approximately 1%). It was found that there were very large differences in ratios in these
sandy sediments, and that these differences could not be related to known sources of
contaminants.
A pragmatic solution was required, and all data with aluminium
concentrations less than 1% were disregarded.
The current advice on normalisation in the Guidelines is based around two alternative
approaches. Firstly the separation and analysis of a defined fine-grained fraction of the
sediment, or secondly the application of geochemical normalisation to analytical data.
Most of the data currently held in the ICES database, and being generated through the UK
NMMP programme, has been obtained through total digestion of either whole (< 2 mm)
sediment, or of a fine-grained fraction (usually < 63 micron). In order to take advantage of
the whole sediment analyses, and to have the possibility to combine them with the < 63
micron data, it is necessary to utilise geochemical normalisation procedures.
The best current advice on this approach is included in the Annex 5 to the OSPAR Sediment
Guidelines, where full details of the procedure are given. In summary, there are two items of
information required to undertake the recommended procedure.
1.
It is necessary to know (or to estimate) the concentrations of both trace element and
the chosen conservative normaliser in a sandy sediment (free from fine grained
material or organic matter) from the area of interest.
2.
It is necessary to obtain expressions of the ratio between the concentrations of the
trace element and the conservative normaliser (say aluminium) in the fine-grained
material in a sediment sample.
The definition and correction for the concentrations of normalisers and contaminants in
sandy sediment should remove the problem encountered by MON assessors in the
interpretation of data from sediments with low aluminium content.
These data can then be handled in one of two ways. If the concentrations of contaminant
and normaliser in sandy sediment (termed the pivot point) are not known, then comparisons
can be made between the gradient of the regression line between trace element and
normaliser and a background value for that ratio in the region of interest.
If the values at the pivot point are known, or can be estimated with sufficient confidence, it is
possible to normalise both background concentrations and concentrations in field samples to
a sediment of “standard” composition, say 5% aluminium. This would correspond broadly to
the composition of muddy sediment.
Having established the basis of the normalisation procedure, it is now appropriate to review
the data available from the UK NMMP and other potential sources of information.
Initial exploration was made of a large data set of analyses of total digestions of whole
(< 2 mm) surface coastal sediment from the Scottish coast. The data were partitioned
geographically into west coast, and a combination of north and east coasts, and the data set
was trimmed to exclude data from estuaries, sea lochs, and also the Clyde sea area. East
and west coast data were treated separately, and linear regression lines were fitted to the
relationships between metal concentrations and aluminium (as conservative normaliser), i.e.
20
Background/Reference Concentrations (BRCs) For The UK
(Concentration of trace metal, ppm) = m*(Concentration of aluminium, %) + c.
The gradients of the linear regressions fitted to the cadmium and mercury data were not
significantly different from zero, i.e. the concentrations of these elements were not
significantly correlated to aluminium. The maximum values for both west and east coast
data for both these elements were approximately 0.2 mg kg-1. As the gradient was not
significantly different from zero, it is not appropriate to use the ratio between these elements
and aluminium as a form of normalisation or as a component of a background concentration.
Rather, the background concentration has expressed in concentration units and
approximates to the maximum concentration of these elements found in sediments remote
for direct sources of contamination.
TABLE 19
Details of significant linear regressions between trace elements and aluminium in Scottish
east coast and North Sea sediments
Standard Error
of m
c
Standard Error
of c
0.553
0.107
2.55
0.24
Pb
2.9
0.25
4.62
0.56
Ni
2.62
0.263
1.19
0.60
Zn
5.9
0.32
3.34
0.72
Cr
9.0
0.44
-2.73
0.99
Element
m (gradient)
As
Insufficient data
Cu
TABLE 20
Details of significant linear regressions between trace elements and aluminium in Scottish
west coast sediments.
Standard Error
of m
c
Standard Error
of c
1.65
0.26
1.32
0.85
Pb
2.9
0.51
4.23
1.66
Ni
5.01
0.69
0.54
2.26
Zn
11.1
1.29
-5.90
4.23
Cr
13.4
1.64
-6.73
5.37
Element
m (gradient)
As
Insufficient data
Cu
The gradients (m) of these regressions therefore provide estimates of the ratio between
trace elements and the normaliser, aluminium. This is one of the two components required
to use the normalisation procedure recommended by OSPAR, and necessary for the
derivation of background values. However, it is apparent that there is significant uncertainty
21
Background/Reference Concentrations (BRCs) For The UK
in the values for the gradients (ratios). The standard errors of the gradient are generally 5 –
20% of the estimated value. This uncertainty will be carried through directly to any
expression of the background value derived from these data.
The other component is the concentrations of both trace element and the chosen
conservative normaliser in a sandy sediment (free from fine grained material or organic
matter) from the area of interest. These concentrations cannot be derived explicitly from the
regression analysis as the true concentration of the normaliser in sand free from fine grained
material or organic matter is not known.
It was suggested at MON 2003 that it might be possible to propose reasonable values for the
concentrations of trace element and the chosen conservative normaliser in sandy sediment
(free from fine grained material or organic matter). These have been determined for The
Netherlands (Smedes, MON 2003), but it is not clear how widely applicable these values
might be within the North Sea, and particularly outside the North Sea. The data currently
being collected through NMMP is targeted at determining temporal trends in concentrations
and therefore the sampling strategy has been, where possible, to avoid areas of sandy
sediment.
The conclusions to be drawn from the above discussion of metals in sediment are:
•
The geochemical normalisation procedures described in the OSPAR Guidelines are
the best currently available, and should be applied when defining
background/reference concentrations in marine sediments.
•
Data currently available in the UK do not allow reliable application of the OSPAR
procedures.
Specifically, the data do not allow reliable definition of the
concentrations of trace elements and normalisers in sandy sediment free from fine
grained material and organic matter.
•
The apparent significant variability in metal to normaliser ratios in sediments from
Scottish coastal waters suggests that these ratios may vary on a more local scale
than the current analysis has examined. This may be reasonable, bearing in mind
the considerable regional differences in geology in the UK, and the known
differences in these ratios between estuaries in England and Wales. It is therefore
recommended that a more detailed assessment is made of the existing data with a
view to defining appropriate divisions of coastal waters within which the inter-element
ratios are less variable and can give a more precise definition of background
conditions.
•
That once these areas have been defined, a specific exercise should be undertaken
to determine the “pivot points” for elements of interest, i.e. a targeted programme of
analysis of sandy sediments is required.
3.3.4
Application of proposed BRC values to NMMP data sets
As part of the process leading to the publication of their latest periodic report, the UK
National Marine Monitoring Programme has compiled and reviewed all the data submitted to
them on contaminants in sediments. The data were filtered for quality, according to the
scheme developed by the Chemical Analytical Quality Control sub-group of the NMMP. The
sediment samples analysed were obtained from England, Scotland, Wales and Northern
Ireland.
22
Background/Reference Concentrations (BRCs) For The UK
As a preliminary exploration of the possible implications of the proposed BRC values for
contaminants in sediments, the final filtered data have been compared to the proposed BRC
values. Tabulations below show the percentages of the filtered data which fall below the
proposed BRC range, within that range, and above that range for PAHs in sediment.
TABLE 21
Application of proposed BRC values to filtered NMMP data on PAHs in sediment
Number
of data
points
Percentage of
data points
below BRC
range
Percentage of
data points
within BRC
range
Percentage of data
points above BRC
range
Naphthalene
314
3
52
45
Phenanthrene
621
3
50
47
Anthracene
641
0
32
68
Fluoranthene
641
3
40
57
Pyrene
640
4
38
58
Benz[a]anthracene
649
0
33
67
Chrysene
647
0
30
70
Benzo[a]pyrene
615
5
41
54
Benzo[ghi]perylene
645
8
52
40
Indenopyrene
572
4
55
41
PAH
23
APPENDIX 1
TECHNICAL ANNEX 5
Normalisation of Contaminant Concentrations in Sediments
1.
Introduction
This annex provides guidance on the application of methods to normalise contaminant
concentrations in sediments. Normalisation is defined here as a procedure to correct
contaminant concentrations for the influence of the natural variability in sediment
composition (grain size, organic matter and mineralogy). Most natural and anthropogenic
substances (metals and organic contaminants) show a much higher affinity to fine particulate
matter compared to the coarse fraction. Constituents such as organic matter and clay
minerals contribute to the affinity to contaminants in this fine material.
Fine material (inorganic and organic) and associated contaminants are preferentially
deposited in areas of low hydrodynamic energy, while in areas of higher energy, fine
particulate matter is mixed with coarser sediment particles which are generally not able to
bind contaminants.
This dilution effect will cause lower and variable contaminant
concentrations in the resulting sediment. Obviously, grain size is one of the most important
factors controlling the distribution of natural and anthropogenic components in sediments. It
is, therefore, essential to normalise for the effects of grain size in order to provide a basis for
meaningful comparisons of the occurrence of substances in sediments of variable
granulometry and texture within individual areas, among areas or over time.
When analysing whole sediment (i.e. < 2 mm fraction) for spatial distribution surveys, the
resulting maps give a direct reflection of the sea bed sediments. However, in areas with
varying grain size distributions, a map of contaminant concentrations will be closely related
to the distribution of fine grained sediments, and any effects of other sources of
contaminants, for example anthropogenic sources, will be at least partly obscured by grain
size differences. Also in temporal trend monitoring, differences in grain size distribution can
obscure trends. If samples used for a spatial survey consist predominantly of fine material,
the influence of grain size distribution is of minor importance and may probably be
neglected.
2.
Normalisation procedures
Two different approaches to correct for variable sediment compositions are widely used:
a)
Normalisation can be performed by relating the contaminant concentration with
components of the sediment that represents its affinity for contaminants, i.e. binding
capacity. Such co-factors are called normalisers (cf. section 4). Normalisation can
be performed by simple contaminant/normaliser ratios or linear regression. Another
procedure takes into account that the coarse sediment fraction contains natural metal
concentrations in the crystal structure before the normalisation is performed (see
section 5). Combinations of co-factors, possibly identified from multiple regression
analysis, can be used as normalisers.
b)
Isolation of the fine fraction by sieving (e.g. <20 µm, <63 µm) can be regarded as a
physical normalisation to reduce the differences in sediment granulometric
compositions and is applicable to both metals and organic contaminants (Ackermann
et al., 1983; Klamer et al., 1990). Consequently the coarse particles, which usually
do not bind anthropogenic contaminants and dilute their concentrations, are removed
from the sample. Then, contaminant concentrations measured in these fine fractions
i
can be directly compared. Subsequently, the differences in sediment composition
due to geochemical nature remaining after sieving can be further corrected for by the
use of co-factors. Thus, sieving is a first powerful step in normalisation.
3.
Limitations of normalisation
Clearly, normalisation procedures may not apply equally well to all elements at all sites;
especially important in this respect are elements that participate in diagenetic reactions. In
cases where there is a lack of full understanding of the geochemical processes operating
care should be taken when normalising for grain size differences. These processes can
create important natural enrichment of metals at the sediment surface, as a result of the
surficial recycling of oxihydroxides or deeper in the sediment as the result of co-precipitation
of the metals with sulphides (cf., e.g., Gobeil et al., 1997), which cannot be accounted for by
normalisation.
There is no evidence that normalised data are more appropriate for ecotoxicological
interpretation than non-normalised data. However, the matter deserves further investigation.
4.
Normalisation with co-factors
a)
The binding capacity of the sediments can be related to the content of fines (primary
factor) in the sediments. Normalisation can be achieved by calculating the
concentration of a contaminant with respect to a specific grain-size fraction such as
<2 µm (clay), <20 µm or <63 µm;
b)
As the content of fines is represented by the contents of major elements of the clay
fraction such as aluminium (Windom et al., 1989) or an appropriate trace element
enriched in that fraction such as lithium (Loring 1991), these can also be used as cofactor (secondary). Both, aluminium and lithium behave conservatively, as they are
not significantly affected by, for instance, the early diagenetic processes and strong
redox effects frequently observed in sediments. Problems may occur in when the
sediment is derived from glacial erosion of igneous rocks, with significant amounts of
aluminium present in feldspar minerals contributing to the coarse fraction. In such
cases, lithium may be preferable (Loring 1991);
c)
Organic matter, usually represented by organic carbon, is the most common cofactor for organic contaminants due to their strong affinity to this sediment
component. Trace metals can be normalised using the organic carbon content
(Cato, 1977) but would require further explanation due to the non-conservative
nature of organic matter.
ii
C
Sand
C ss
Cs
Slope PL=dC/dN
Cx
Co-factor i.e.
normaliser
Nx
Figure 1.
5.
Ns
N ss
Relation between the contaminant C and the cofactor N (see text).
Theory
The general model for normalisation taking into account the possible presence of
contaminants and cofactors in the coarse material is given in figure 1 (Smedes et al., 1997).
Cx and Nx represent the co-factor and the contaminant contents, respectively, in pure sand.
These “intercepts” can be estimated from samples without fines and organic material. The
line of regression between the contaminant and co-factor will originate from that point. That
means that regression lines of sample sets with a different pollution level and consequently
different slopes will have this point in common (i.e. pivot point). When this pivot point is
known only one sample is required to estimate the slope. This allows determination of the
contaminant content for any agreed (preselected) co-factor content (Nss) by interpolation or
extrapolation. The slope for a sample with a contaminant content Cs and a cofactor content
of Ns can be expressed as follows:
PL =
dC C s - C x
=
dN N s − N x
(1)
The extrapolation to an agreed co-factor content, Nss, follows the same slope:
PL =
dC C s - C x
=
= C ss C x
dN N s − N x N ss − N x
(2)
Rewriting gives the contaminant content, Css, that is normalised to Nss:
C ss = ( C s - C x )
N ss - N x + C
x
Ns - Nx
(3)
Results of different samples normalised to the agreed Nss can be compared directly.
Normalisation by this model can be applied with different cofactors. Here primary and
secondary cofactors can be distinguished. A primary cofactor like clay or organic carbon is
not present in the coarse fraction and consequently has no intercept (Nx=0). Al and Li are
iii
present in the coarse fraction and therefore are considered to be secondary cofactors.
Provided Nx and Cx are known, the model allows recalculation of total samples to a co-factor
content usually found in sieved fractions, either <20 or <63 µm. However such an
extrapolation for a coarse grained sample will be associated with a large error due to the
uncertainty of the intercepts and the analysed parameters. For a more fine grained sample,
the uncertainty of the normalised result is much lower than for normalisation of a sieved
fraction to the agreed cofactor content and will result in a more accurate result. The model
presented also applies to the normalisation of organic contaminants using organic carbon
but in that case the intercepts Nx and Cx will not differ significantly from zero.
Principally, the result allows comparison of data of total and sieved samples, irrespective the
sieving diameter but the error has to be taken into account. Through propagation of errors
the standard error of the result can be calculated from the analytical variation and the natural
variation of the intercept Nx. Results can therefore always be reported with a standard
deviation.
6.
Considerations on co-factors
The clay mineral content is the most important cofactor for trace metals. In the model
above the Nx will be zero for clay and only the intercept due to the content of the trace metal
in the coarse fraction (Cx) has to be taken into account. However, current intercomparison
exercises do not include this parameter. Presently other parameters such as aluminium or
lithium are used to represent the clay content.
The aluminium content in the sandy fraction may vary from area to area. For some areas
aluminium contents in the sandy fractions are found at the same level as found in the fines
(Loring, 1991) and therefore the intercept Nx becomes very high. In equation (3) this implies
that the denominator is the result of subtracting two large numbers, that is the normaliser
content in the sample (Ns) and the normaliser content in only sand (Nx). Consequently, due
to their individual uncertainties, the result has an extremely high error. Obviously,
normalisation with low intercepts is more accurate. Much lower intercepts are found if partial
digestion methods are used that digest the clay minerals, but not the coarse minerals. Using
partial digestion, the spatial variability of the results of aluminium analyses in the sandy
fraction has been found to be much smaller than with total methods. Although normalising
concentrations of contaminants in fine grained material will always give more accurate
results, an error calculation will identify whether using coarse samples (and total methods,
e.g. HF, X-ray fluorescence) allows the requirements of the program to be met.
For most areas the lithium content in the sandy fraction is much lower than in the fine
fraction. In addition, results from partial digestion and total methods do not differ
significantly. There is only little spatial variability of the lithium content in the sandy fraction.
Generally, compared to aluminium, more accurate normalised data can be expected using
lithium.
As for clay, no intercept (Nx) applies for organic matter, which is usually represented by
organic carbon. Organic matter also occurs in the coarse fraction but is even then a
cofactor that contributes to the affinity for contaminants, whereas the aluminium in the
coarse fraction does not. Furthermore, organic matter in a sample is not always well defined
as it can be composed of material with different properties. The most variable properties will
be found in the organic matter present in the coarse fraction, i.e. that not associated with the
fines. In fine sediments or in the sieved fine fractions the majority of the organic matter is
associated with the mineral particles and it is assumed to be of more constant composition
than in the total sample. In addition, the nature of the organic matter may show spatial
variation. For samples with low organic carbon content close to the detection limit,
normalisation using this cofactor suffers from a large relative error. This results from the
iv
detection limit and the insufficient homogeneity that cannot be improved due to the limited
intake mass for analysis.
For further interpretation of data the proportion of fines determined by sieving can be
useful. Provided, there are no significant amounts of organic matter in coarse fractions, the
proportion can be used as normaliser. The error in the determination of fines has to be
taken into account and will be relatively high for coarse samples.
7.
Considerations on contaminants
Almost all trace metals, except mercury and in general also cadmium, are present in the
coarse mineral matrix of samples. The metal concentrations show a spatial variability
depending on the origin of the sandy material. In sandy sediments, partial digestion
techniques result in lower values than are obtained from total digestion techniques. This
implies that partial digestion results in lower intercepts (pivot point is closer to the zero).
However, the partial digestion must be strong enough so the clay will be totally digested (as
is the case with HF digestion techniques), and the measured aluminium content remains
representative for the clay. It was demonstrated that analyses of fine material gave similar
results for several trace elements using both total and strong partial methods (Smedes et al.
2000, QUASH/QUASIMEME intercalibrations).
In general, correlations of organic contaminants with organic carbon have no significant
intercept. Obviously a normalised result from a coarse sample will show a large error as due
to the dilution by sand the concentrations are often close or even below the detection limit.
Presently, organic carbon is usually applied for normalisation of PAHs. It should be
recognised that due to the possible presence of undefined material, for example soot or ash,
elevated PAH concentrations may occur in specific fractions that might have limited
environmental significance. Although this needs further investigation, existing results
indicate that PAH concentrations in the sieved fractions are not affected significantly.
8.
Isolation of fine fractions for analyses
The Sample Preparation
Samples must be sieved at 2 mm as soon as possible after sampling to remove large
detritus and benthic organisms. Otherwise during further sample handling like storage,
freezing or ultrasonic treatment, biotic material will deteriorate and become part of the
sediment sample. Until the final sieving procedure that isolates the fines, the sample can be
stored at 4°C for about a week and up to 3 months when frozen at –20°C, although direct
wet sieving is preferred. For prolonged storage freeze-drying of samples can be considered.
In this case contamination and losses of contaminants during freeze-drying have to be
checked. Air-drying is not appropriate due to high contamination risks. Besides, samples
may be difficult to be disaggregate and mineral structures may be affected.
Requirements for Sieving
A wet sieving procedure is required to isolate the fine-grained fractions (<63 µm or <20 µm).
Wet sieving re-suspends fine particles that would otherwise remain attached to coarser
particles in the sample. Sediments should be agitated during sieving to prevent to
disaggregate agglomerates of fines and to prevent clogging of the mesh. Freeze-dried
samples need to be re-suspended using ultrasonic treatment. Seawater, preferably from the
sampling site, should be used for sieving as it reduces the risk of physico-chemical changes
in the sample i.e. losses through leaching or contamination. Furthermore seawater assists
the settling of fine particles after the sieving. If water from the sampling site is not available,
then seawater of an unpolluted site, diluted with deionised water to the required salinity, can
v
be used. The amount of water used for sieving should be kept to a minimum and be reused
for sieving subsequent batches.
To minimise or prevent contamination it is recommended to use large sample amounts of
sediment for sieving. No significant contaminant losses or contamination was detected
when at least 25 g of fine fraction is isolated. (QUASH).
Methodology
Both automated and manual methods are available for sieving. A video presentation of
these methods can be provided by the QUASH Project (QUASH 1999).
•
The automatic sieving method pumps seawater over a sieve that is clamped on a
vibrating table (Klamer et al., 1990). The water passing the sieve is lead to a flowthrough centrifuge that retains the sieved particles and the effluent of the centrifuge is
returned to the sieve by a peristaltic pump. Large sample amounts, up to 500 g, can
be handled easily.
•
The second method is a manual system sieving small portions 20-60 g using an 8 cm
sieve in a glass beaker placed in an ultrasonic bath (Ackermann et al., 1983).
Particles are isolated from the water passing the sieve by batch wise centrifugation.
The water can be reused for a subsequent batch of sediment. In case of sandy
samples, when large amounts of sediments have to be sieved, removal of the coarse
material by a pre-sieving over e.g. 200 µm mesh can facilitate the sieving process.
Isolated fine fractions have to be homogenised thoroughly, preferably by a ball mill, as
centrifugation produces inhomogeneous samples due to differences in settling speed of
different grain-size fractions.
9.
Recommendations
1.
For both temporal trend and spatial monitoring, it would be ideal to analyse samples
with equal composition. This could be confirmed by determination of co-factors Al,
Li, OC and parameters of the grain size distribution (e.g. clay content, proportion <20
µm, proportion <63 µm). However, this situation will not always occur, particularly in
the case of spatial surveys.
2.
New temporal trend programs should be carried out by the analysis of fine sediments
or a fine-grained fraction, isolated by sieving. Existing temporal trend programs
could be continued using existing procedures, provided that assessment of the data
indicates that the statistical power of the programs is adequate for the overall
objectives.
3.
Contaminant concentrations in whole sediments can be subjected to normalisation
using co-factors for organic matter, clay minerals etc., taking into account the
presence of both co-factors and target contaminants in the mineral structure of the
sand fraction of the sediment. Taking into account these non-zero intercepts of
regressions of contaminant concentrations with co-factors, normalisation to
preselected co-factor content will reduce the variance arising from different grain
sizes. Normalised values for sandy sediments will have greater uncertainties than for
muddy sediments. The propagated error of the variables used for normalisation may
be unacceptable high for sandy sediments, if both contaminant and co-factor
concentrations are low, particularly when approaching detection limits. In that case,
in order to draw reliable maps, alternative procedures, such as sieving, need to be
used to minimise the impact of this error structure.
vi
4.
Variance arising from grain size differences can be reduced in a direct way by
separation of a fine fraction from the whole sediment. Spatial distribution surveys of
the concentrations of contaminants in separated fine fractions can be used to
prepare maps which will be much less influenced by grain size differences than maps
of whole sediment analyses. There will still be some residual variance arising from
differences in the composition (mineralogy and organic carbon content) of the
sediments.
5.
The natural variance of sample composition will be smaller in the fraction <20 µm
than in the fraction <63 µm. Therefore, the fraction <20 µm should be preferred over
the fraction <63 µm. However, separation of the fraction <20 µm can be
considerably more laborious than the separation of the fraction <63 µm and might be
an obstacle to its wide application. For this practical reason, the fraction <63 µm is
an acceptable compromise for both temporal trend and coordinated large scale
spatial surveys.
6.
The preferred approach for preparing maps of the spatial distribution of contaminants
in sediment consists of two steps: analyses of contaminants in fine sediments or in
the fraction <63 µm, followed by normalisation of analytical results using co-factors
(see section 4). Current scientific knowledge indicates that this procedure minimises
the variances arising from differences in grain size, mineralogy and organic matter
content. Application of this two-tiered approach to fractions <20 µm gives results that
can be directly compared to results found by normalisation of concentrations
measured in fractions <63 µm.
This approach should give consistent and
comparable data sets over the ICES/OSPAR area. Maps of contaminant levels in
fine sediments should be accompanied by maps of the co-factors in the whole
sediments.
In order to clarify aspects of data interpretation, analytical data for field samples should be
accompanied by information on limits of detection and long term precision. In order to
contribute to environmental assessment, data for field samples should include the grain size
distribution, as a minimum the proportion of the analysed fraction in the original whole
sediment.
10.
References
Ackermann, F., Bergmann, H. and Schleichert, U. 1983. Monitoring of heavy metals in
coastal and estuarine sediments - A question of grain-size: <20 µm versus <60 µm.
Environmental Technology Letters, 4, 317-328.
Cato, I. 1977. Recent sedimentological and geochemical conditions and pollution problems
in two marine areas in south-western Sweden. Striae 6, 158 pp. Uppsala.
Gobeil, C., MacDonald, R.W. and Sundby, B. 1997. Diagenetic separation of cadmium and
manganese in suboxic continental margin sediments. Geochim. Cosmochim. Acta,
61, 4647-4654.
Klamer, J.C., Hegeman, W.J.M. and Smedes, F. 1990. Comparison of grain size correction
procedures for organic micropollutants and heavy metals in marine sediments.
Hydrobiologia, 208, 213-220.
Loring, D.H. 1991. Normalization of heavy-metal data from estuarine and coastal
sediments. ICES J. Mar. Sci., 48, 101-115.
vii
QUASH. 1999. Sediment Sieving Techniques, QUASH Project Office, FRS Marine
Laboratory, PO Box 101, Victoria Road, Aberdeen, AB11 9DB, Scotland.
Smedes, F. 1997. Grainsize Correction Procedures, Report of the ICES Working Group on
Marine Sediments in Relation to Pollution. ICES CM 1997/Env:4, Ref E, Annex 6.
Smedes, F., Lourens, J. and Wezel, van A. 1997. “Zand, Slib en Zeven, Standardisation of
contaminant contents in marine sediments, Report RIKZ-96.043 (Dutch), ISSN 09273980, RIKZ, PO Box 20907, 2500 EX, The Hague.
Smedes, F., Davies, I.M., Wells, D., Allan, A. and Besada, V. 2000. Quality Assurance of
Sampling and Sample Handling (QUASH) - Interlaboratory study on sieving and
normalisation of geographically different sediments; QUASH round 5 (sponsored by
the EU Standards, Measurements and Testing Programme)
Windom, H.L. et al. 1989. Natural trace metal concentrations in estuarine and coastal
marine sediments of the southeastern United States. Environ. Sci. Technol., 23,
314-320.
viii