Date post: | 21-Nov-2023 |
Category: |
Documents |
Upload: | independent |
View: | 0 times |
Download: | 0 times |
Environmental Pollution 148 (2007) 73e82www.elsevier.com/locate/envpol
Modelling PCB bioaccumulation in a Baltic food web
Erick Nfon, Ian T. Cousins*
Department of Applied Environmental Science (ITM), Unit for Environmental Toxicology and Environmental Chemistry,
Frescativagen 50, Stockholm University, SE 10691, Stockholm, Sweden
Received 31 March 2006; received in revised form 19 October 2006; accepted 1 November 2006
The bioaccumulation behaviour of PCB congeners in a Baltic food web is studied using a novel mechanistic model.
Abstract
A steady state model is developed to describe the bioaccumulation of organic contaminants by 14 species in a Baltic food web includingpelagic and benthic aquatic organisms. The model is used to study the bioaccumulation of five PCB congeners of different chlorination levels.The model predictions are evaluated against monitoring data for five of the species in the food web. Predicted concentrations are on averagewithin a factor of two of measured concentrations. The model shows that all PCB congeners were biomagnified in the food web, which is con-sistent with observations. Sensitivity analysis reveals that the single most sensitive parameter is log KOW. The most sensitive environmental pa-rameter is the annual average temperature. Although not identified amongst the most sensitive input parameters, the dissolved concentration inwater is believed to be important because of the uncertainty in its determination. The most sensitive organism-specific input parameters are thefractional respiration of species from the water column and sediment pore water, which are also difficult to determine. Parameters such as feed-ing rate, growth rate and lipid content of organism are only important at higher trophic levels.� 2007 Elsevier Ltd. All rights reserved.
Keywords: Food web; Model; PCBs; Baltic Sea; Bioaccumulation
1. Introduction
The overall behaviour of pollutants released into theenvironment may be assessed by structuring the environmentinto different compartments, developing mathematicalrelationships to describe the fate and behaviour in a compart-ment and transport from one compartment to another (Mackay,2001). Similarly, food web models have been developed to de-scribe the uptake and bioaccumulation of organic pollutants bysingle organisms and in aquatic food webs (Neely et al., 1974;Clark et al., 1990; Clark and Mackay, 1991; Thomann et al.,1992; Gobas, 1993; Morrison et al., 1996; Campfens andMackay, 1997; Morrison et al., 1997; Endicott et al., 1998;Fraser et al., 2002; Czub and Mclachlan, 2004). The primaryconcern in food web models is the phenomenon by which
* Corresponding author. Tel.: þ46 8 16 4012; fax: þ46 8 674 7638.
E-mail address: [email protected] (I.T. Cousins).
0269-7491/$ - see front matter � 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.envpol.2006.11.033
pollutants present at low concentrations in water becomeconcentrated by many orders of magnitude in fish, birds andhumans who consume fish (Mackay, 2001; Kelly et al.,2004). The uptake of pollutants by aquatic organisms occursvia water (by gills, epidermis) or diet, however, dietary expo-sure is usually the dominant pathway of uptake for organismsat higher trophic levels in aquatic and terrestrial food webs(Thomann and Connolly, 1984; Clark et al., 1990; Gobaset al., 1993; Sharpe and Mackay, 2000).
The polychlorinated biphenyls (PCBs) have emerged as im-portant pollutants of concern because of their ubiquitouscharacter (Kjeller and Rappe, 1995; Roots and Talvari, 1997;Bignert et al., 1998; Nyman et al., 2002) the tendency tobioaccumulate within food webs from water and sediment toaquatic invertebrates (Koistinen et al., 1995; Strandberget al., 1998; Kiviranta et al., 2003) and their relative toxicity(Konat and Kowalewska, 2001). The Baltic Sea is particularlyvulnerable to contamination by organic contaminants due tothe low diversity of species and slow water exchange with
74 E. Nfon, I.T. Cousins / Environmental Pollution 148 (2007) 73e82
the open ocean. Many studies have revealed the presence ofPCBs in air, water and sediment samples collected from theBaltic region (Kjeller and Rappe, 1995; Bignert et al., 1998;Jonsson and Carman, 2000; Kiviranta et al., 2003).
The objectives of this study were: (a) to develop a steadystate food web model with a capability to predict PCB levelsand assess the importance of the different uptake and elimina-tion processes by 14 organisms in a Baltic food web; (b) toevaluate the model performance by comparing modelpredicted concentrations in organisms to measured concentra-tions; and (c) identify by sensitivity analysis the inputparameters that significantly influenced the variance in the pre-dicted concentrations. Previous papers on the concentrationsand bioaccumulation of organic pollutants in the Baltic havefocussed on the short food chains characteristic of the Baltic(Rolff et al., 1995; Strandberg et al., 1998; Burreau et al.,2004). Furthermore, the only known mechanistic food webmodel available for the Baltic consists of a short marine sys-tem that includes three pelagic species namely zooplankton,planktivorous and piscivorous fish and considers the differentlife stages of these species as the same organism (Czub andMclachlan, 2004). This study presents a novel approach inBaltic food web modelling due to the following reasons. First,to the best of our knowledge, this is the most extensive mech-anistic food web model developed for the Baltic and includesseveral trophic levels comprising pelagic and benthic aquaticorganisms. Secondly, the model treats the different life stagesof one species as different organisms, which was deemedappropriate since the different life stages of the same speciesshow variability in their dependence on a particular prey anddiverse physiological characteristics (Harvey et al., 2003;Gorokhova et al., 2004). For example, the diet of juvenile her-ring consists of 90% mezozooplankon, 6% pelagic macrofaunaand 3% benthic macro fauna while the diet of adult herringconsists of 70% mezozooplankon, 6% pelagic macrofauna,8% juvenile sprat and 11% juvenile cod (see Table 3).
2. Methods
2.1. Model development
The Baltic food web model developed in this study comprises 14 target
‘‘organisms’’ representing different trophic levels and guilds in the Baltic
Sea, namely: bacteria, phytoplankton (e.g. diatoms cyanobacteria and dinofla-
gellates), microzooplankton (e.g. Acartia spp.), mezozooplankton (e.g. Euryte-
mora spp.), pelagic macro fauna (Mysis sp.), benthic meiofauna (e.g. ostracods
and harpacticoid copepods), benthic macro fauna (e.g. amphipod e Monopor-eia affinis and isopod e Saduria entomon), juvenile sprat (Sprattus sprattus),
juvenile herring (Clupea harengus), juvenile cod (Gadus morhua), adult sprat,
adult herring, adult cod and salmon (Salmo salar). Defining guilds rather than
specific organisms was preferred at lower trophic levels because there are
multiple species within these guilds.
The model is based on the approach of Campfens and Mackay (1997) and
describes the uptake (by respiration and dietary transfer) of PCBs by
organisms in a Baltic food web and uses the fugacity concept to express these
distributions. Since the formulation of the model is based on a previously
developed model we do not include full descriptions of model equations in
the text. We refer readers interested in the model details to the supplementary
information available for this article. A brief description is given below.
The steady state fugacity, fF (Pa), of each organism in the food web is
expressed by Eq. (1) (Campfens and Mackay, 1997)
fF ¼ fWW þ fAA ð1Þ
where W and A are fugacity factors for respiration or diffusive uptake from
water and food, fW (Pa) is the fugacity of water, and fA (Pa) is the fugacity
of food. Pelagic species respire only in the water column, while benthic organ-
isms respire both in the water column and sediment pore water. Eq. (1) is thus
modified as follows
fF ¼WðxWfW þ xS fSÞ þAfA ð2Þ
The parameters xW and xS are the fractional respiration from water and
pore water, respectively.
Following the approach of Campfens and Mackay (1997), a mass balance
equation was written for each organism in the food web and then these equa-
tions linked together to describe transport and transformation in the whole
food web. The only exception was for estimation of concentrations in bacteria
and phytoplankton which were calculated using an equilibrium partitioning
equation (Morrison et al., 1997)
CF ¼ CWKOC fOC ð3Þ
where CW (g/l) is the freely dissolved water concentration, KOC is the organic
carbon water partitioning coefficient and fOC (g/g) is the organic carbon con-
tent of bacteria and phytoplankton.
Following the approach of Campfens and Mackay (1997) and Sharpe and
Mackay (2000), the general expression describing the total uptake (by respira-
tion and through the diet) for the 14 organisms was written as
Af ¼ E ð4Þ
where A is a (14� 14) biomagnification matrix, f is a vector of organism fugac-
ities ( f1 to f14) and E is a respiration vector characterizing the fugacity induced
in the organism from its abiotic environment with elements of the form
WiðxWi fW þ xSi fSÞ For i¼ 1e14 ð5Þ
A steady state solution for this equation was generated as described in
Campfens and Mackay (1997) and Sharpe and Mackay (2000). Steady state
in this context refers to the following assumptions generally used in steady
state modelling approaches (Mackay and Fraser, 2000), i.e. (i) growth rates
of invertebrates and fish species are linear and all processes are first order
with respect to chemical concentration; and (ii) the organism has maintained
constant D-values for a period of time which is long relative to the clearance
time of the contaminant. Furthermore, the model is based on the assumption
that internal equilibrium exists within organisms and predicted concentrations
are for whole body weight. The use of a steady state introduces a simplifying
assumption since there are temporal differences in contaminant levels (Nfon
and Cousins, 2006) and thus exposure. Since the steady state approach has
been shown to be consistently successful for a range of other aquatic ecosys-
tems (e.g. Campfens and Mackay, 1997; Morrison et al., 1997) it appears that
changes in PCB levels are sufficiently slow so that steady state is a reasonable
approximation. It is important, however, that input abiotic concentrations are
from the same time period as the biotic concentrations used for model
evaluation.
2.2. Model parameterisation
2.2.1. Physicalechemical properties and metabolism rates of PCBs
Selecting reliable physicalechemical properties is extremely important in
environmental modelling due to the relative significance of physicalechemical
properties with respect to partitioning and mass transfer of chemicals (for ex-
ample, Beyer et al., 2002; Li et al., 2003 for a review). Five PCB congeners
(PCB 28, PCB 101, PCB 138, PCB 153 and PCB 180) covering a range of
chlorination were selected for modelling. Physicalechemical properties of
PCBs used as model inputs were taken from a consistent data set in Mackay
et al. (1992) and Li et al. (2003). These sources were regarded as reliable since
the physicalechemical properties were generated from a larger data set that
had undergone the required adjustments and quality checks for consistency.
75E. Nfon, I.T. Cousins / Environmental Pollution 148 (2007) 73e82
Physicalechemical properties were corrected to reflect the typical
conditions of the Baltic assuming an average annual temperature of 7 �C for
the Baltic (Sinkkonen and Paasivirta, 2000). The KOW was corrected using
the Van’t Hoff expression; for the sake of consistency, the Henry’s law constant
H was calculated from the KOW and the KOA (both corrected for temperature,
see Table 1 for details). The KOC was determined as 0.41 KOW (Mackay,
2001), and the change of KOC was thus proportional to change in KOW. The ef-
fect of salinity on the tendency of PCBs to partition into organic phases was
accounted for via the Setschenow constant ðKsi Þ assuming a salinity of 6&
for the Baltic Sea (Rodhe and Winsor, 2002). A mean value of 0.035 M�1
for Ksi was determined for PCB congeners using literature citations (Schwar-
zenbach et al., 2003).
Generally, the metabolism rate constant for PCB congeners by aquatic
species is expected to be both congener and species dependent, hence the
complete parameterisation for the present model would require data for the
metabolism of five PCB congeners by 14 species. Due to the lack of empirical
data on metabolic transformation in PCBs as has been reported by other au-
thors (for example, Arnot and Gobas, 2003), we assumed an infinite half-
life (t1/2) for PCB congers. This makes the metabolism rate constant kM
(kM¼ ln 2/t1/2) negligible and the D-value for metabolism insignificant. The
implication of this assumption is that it might lead to an overestimation of pre-
dicted concentrations in species were metabolism is significant.
2.2.2. Environmental parameters for the Baltic
Eight input parameters were used to describe the environmental character-
istics of the Baltic. Included were suspended particulate matter concentration,
organic carbon fraction of suspended matter and sediment particulates, the
volume fraction of sediment particulates and the density of particulates in
the water column. Additional model inputs were the dissolved water concen-
tration of each PCB congener, the chemical sediment concentration and tem-
perature (discussed above); input values for these parameters were taken from
the literature and are listed in Table 2.
The water and sediment concentrations in Table 2 were converted to
fugacities by the model. The fugacity in water is derived from measured
dissolved water concentrations. These concentrations may not be the truly
dissolved concentration as there may be some partitioning to dissolved
organic matter (DOM) or colloidal material (Meylan et al., 1999; Konat and
Kowalewska, 2001; Borga et al., 2004). We decided not to use estimation
techniques that account for DOM partitioning to estimate the truly dissolved
concentration (Gschwend and Wu, 1985). The sensitivity of the dissolved water
concentrations on model predicted concentrations in organisms is discussed
later.
2.2.3. Parameters describing food web species
The weight of fish species were determined from generalized fish data
(ICES, 2001). Masses were not required for bacteria and phytoplankton as
equilibrium partitioning was assumed (Eq. (3)). The precise mass and volume
of small organisms such as zooplankton is uncertain and is only useful in de-
termining the time required to attain equilibrium, which we believe is short.
We have therefore input an arbitrary low mass for zooplankton (Table 2). Sen-
sitivity analysis later revealed that variation of this mass had negligible impact
on predicted concentrations.
The feeding rates and growth rates are from Hansson (Personal communi-
cation). The dietary preference of 13 of the 14 organisms were taken from
a previously developed ecosystem model for the Baltic (Harvey et al.,
2003). The different prey items in the diet of salmon were derived directly
from data reported in Hansson et al. (2001). The dietary preferences of the
organisms are listed in Table 3 and all other organism-specific inputs are listed
in Table 4. The mathematical expressions used to determine specific organism
parameters are presented in the supplementary information.
The values of xW and xS in Eq. (2) were set as one and zero, respectively,
for all fish species respiring exclusively in the water column. It proved difficult
to estimate xW and xS for the benthic species since they show a variation in
habitat depending on, for example, diurnal conditions and the presence or
absence of a predator (Hill, 1991; Ejdung, 1998; Stevenson, 2003). Benthic
organisms bioturbate sediments to enhance water circulation and oxygenation,
however, due to the anoxic nature of bottom sediments, sediment dwelling or-
ganisms that need oxygen either live close to the surface or maintain a burrow
to allow water circulation and oxygenation (Snelgrove, 1999). A review of the
literature revealed that various authors had used different approximations to
describe the fraction of pore water respiration of benthic organisms. For
example, Campfens and Mackay (1997) assumed 100% pore water respiration
Table 2
Environmental input parameters for the Baltic
Input parameter Unit Input value Reference
Annual average water
temperature
�C 7 Sinkkonen and
Paasivirta (2000)
Suspended particulate
matter concentration
g/m3 4 Granskog (1999)
Organic carbon fraction
of suspended matter
0.05 Hakanson et al. (2004)
Organic carbon fraction
of sediment particles
0.05 Huttig and Oehme (2005)
Volume fraction of
sediment particles
0.3 Huttig and Oehme (2005)
Density of particles
in water column
kg/m3 2500 Mackay (2001)
Input concentrations
of PCB congeners
CW (ng/l) CS (ng/g)
PCB 28 0.0009 5.2
PCB 101 0.0020 8.6
PCB 138 0.0006 2.6
PCB 153 0.0005 3.1
PCB 180 0.0001 2.2
CW, Input concentration in water; CS, input sediment concentration.
Table 1
Physicalechemical properties of PCBs that constitute model input
PCB 28 PCB 101 PCB 138 PCB 153 PCB 180 References
MW (g/mol) 257.54 326.4 360.9 360.9 395.32 Mackay et al. (1992)
log KOW 5.67 6.3 7.21 6.87 7.16 Li et al. (2003)
log KOA 7.85 8.73 9.66 9.44 10.16 Li et al. (2003)
H (Pa m3/mol) 30.5 24.1 30.1 19.8 8.1
log KOWa 5.47 6.14 7.03 6.68 6.97
log KOAa 7.85 8.73 9.66 9.44 10.16
DOW KJ/mol �26.3 �23.8 �25.0 �31.1 29.1 Li et al. (2003)
DOA KJ/mol �78.5 �83.5 �86.3 �93.9 �92.8 Li et al. (2003)
DAW KJ/mol 52.3 59.7 61.3 62.8 63.6 Li et al. (2003)
MW is the molecular weight, H is the Henry’s law constant, KOW and KOA are the octanolewater, octanoleair and the airewater partition coefficients, respectively.
H was estimated as KAW�RT (KAW was calculated from KOW/KOA). DOW, DOA and DAW are the internal energies of phase transfer.a Corrected for temperature and salinity.
76 E. Nfon, I.T. Cousins / Environmental Pollution 148 (2007) 73e82
Table 3
Dietary preference matrix and trophic positions (TP) for organisms
Prey
TP Predator De Ba Phyt Mi.Z Me.Z P.Ma B.Me B.Ma J.sprat J.He J.Cod A.sprat A.He A.Cod Sa
1 Ba 1
1 Phyt
2 Mi.Z 0.79 0.21
2.25 Me.Z 0.75 0.25
2.6 P.Ma 0.5 0.5
3.5 B.Me 1
2.8 B.Ma 0.67 0.04 0.0009 0.0001
3.2 J.sprat 0.999
3.3 J.He 0.9 0.06 0.03 0.14
3.6 J.Cod 0 0.47 0.0009 0.230 0.14 0.02
3.2 A.sprat 0.999 0.001
4.1 A.He 0.87 0.1 0.03 0.102 0.138
4.2 A.Cod 0.219 0.14 0.296 0.07 0.166 0.079
4.3 Sa 0.59 0.25 0.13 0.03
De, Detritus; Ba, bacteria; Phyt, phytoplankton; Mi.Z, mizozooplankton; Me.Z, mezozooplankton; P.Ma, pelagic macrofauna; B.Me benthic meiofauna; B.Ma,
benthic macrofauna; J.sprat, juvenile sprat; J.He, juvenile herring; J.Cod, juvenile cod; A.sprat, adult sprat; A.He, adult herring; A.Cod, adult cod; Sa, salmon.
for Lake Ontario benthic species, Winsor et al. (1990) assumed a 4% pore
water respiration for Macoma nasuta while Arnot and Gobas (2004) approxi-
mated a 5% pore water respiration for benthic species in Lake Erie and Lake
St. Claire. For ‘‘benthic’’ organisms included in the present model, we
assumed a 10% sediment pore water respiration (i.e. xS¼ 0.1 and xW¼ 0.9)
after consultation with scientists who study benthic organisms (Prof. Dag
Broman and Assoc. Prof. Brita Sundelin, ITM, Stockholm University, personal
communication). We admit that this value is very uncertain, but the fraction
used is our best estimate. It will be replaced if better data become available.
The sensitivity of the model to this input parameter is examined in the
sensitivity analysis presented later in this paper.
2.3. Sorting organisms into trophic position
In order to aid presentation and interpretation of model predictions,
organisms were sorted into approximate tropic position following an approach
outlined in Mackintosh et al. (2004). The estimated trophic positions of the
organisms are listed in Table 3. Abiotic media (air and water) together with
phytoplankton were assigned a default trophic position of one and sediment
was assigned a value of 2.5. Using the derived trophic levels, we classified
the organisms in the food web into four trophic levels as follows; level 1 (phy-
toplankton, bacteria); level 2 (microzooplankton, mezozooplankton, pelagic
macrofauna, benthic macrofauna) level 3 (juvenile sprat, adult sprat, juvenile
Table 4
Organism characteristics for model parameterisation
Species M (g) LF GIPV (g/gd) GRRD (g/gd) xW xS QD GV (L/d) EW
Bacteria e 0.25d 0.678e 0.391e 1 3h 0.002i 0.9i
Phytoplankton e 0.25d 0.192e 1 3h 0.002i 0.9i
Microzooplankton e 0.015e 1.487e 0.587e 1 3h 0.002i 0.9i
Mezozooplankton e 0.015e 0.822e 0.226e 1 3h 0.002i 0.9i
Pelagic macrofauna 0.012a 0.07e 0.068e 0.021e 1 3h 0.002i 0.9i
Benthic meiofauna 0.0052b 0.52f 0.085e 0.017e 0.90 0.10g 3h 0.5i 0.9i
Benthic macrofauna 0.01b 0.52f 0.036e 0.001e 0.90 0.10g 3h 0.5i 0.9i
Juvenile sprat 7.9c 0.04e 0.058e 0.002e 1 3h
Juvenile herring 17.6c 0.04e 0.040e 0.001e 1 3h
Juvenile cod 160c 0.05e 0.007e 0.001e 1 3h
Adult sprat 12.3c 0.04e 0.028e 0.002e 1 3h
Adult herring 33.2c 0.044e 0.022e 0.001e 1 3h
Adult cod 3944c 0.055e 0.005e 0.003e 1 3h
Salmon 4700c 0.16e 0.032e 0.002e 1 3h
M, Mass of organism in grams; LF, lipid fraction; GIPV, feeding rate as percent of body per day; GRRD, growth rate as fraction of volume per day; xW, fractional
respiration from water; xS, fractional respiration from pore water; EA, gut absorption efficiency; QD, maximum biomagnification factor; QW, water transport pa-
rameter (L/d); GV, gill uptake rate (L/d); EW, gill uptake efficiency.a Broman et al. (1992).b Breitholtz et al. (2001).c ICES (2001).d Granskog (1999).e From Hansson (unpublished data).f Strandberg et al. (1998).g Estimated (personal communication, see text).h Campfens and Mackay (1997).i Gobas and Wilcockson (2003).
77E. Nfon, I.T. Cousins / Environmental Pollution 148 (2007) 73e82
herring and benthic meiofauna); level 4 (juvenile cod, adult herring, adult cod,
salmon). Trophic positions differed within a trophic level, for example within
trophic level 3; juvenile herring was at a higher trophic position than juvenile
sprat whereas within level 4, salmon was at a higher position than adult cod.
2.4. Monitoring data used for model evaluation
The model development and evaluation exercise required two monitoring
data sets of PCB congener concentrations, an abiotic data set for model
simulation and a biotic (evaluation) data set for evaluation of the model
predictions. The abiotic data set included the dissolved PCB concentration in
water and sediments and the biotic (evaluation) data set contained monitoring
data of PCB concentrations in food web species.
Both the abiotic and biotic data sets were taken from a previously devel-
oped database of organic pollutants for the Baltic Sea. Nfon and Cousins
(2006) described this database in detail so it is only briefly outlined here.
Concentrations of persistent organic pollutants, including PCBs, in abiotic
and biotic samples from a variety of data sources were compiled in a database.
The database covered samples collected from multiple locations in the Baltic,
representing background sites over a 32 years period (1970e2002). However,
the concentrations used as model input and model evaluation were a subset of
data for the period 1990e2000.
Abiotic samples were collected at different locations, depths and reported
in a variety of different units. In some cases abiotic data were reported as total
concentrations, in others dissolved and particulate concentrations and in some
cases the speciation was unspecified. Biotic samples including benthic inver-
tebrates and fish species of different sizes and ages were also included in
the database.
In generating our simulation and evaluation data sets, the following cri-
teria were used to eliminate any sources of variability and errors: (a) the con-
centrations used as model input and the data used to compare the model
predictions represent data for 1990e2000; (b) only water concentrations re-
ported as dissolved were used; and (c) care was taken to include only data of
congener concentrations in fish species that could be separated into juveniles
or adults. A general criterion of the database of Nfon and Cousins (2006)
was that only data from background sites, away from obvious point sources,
be included.
Predicted and measured biota concentrations were lipid normalized by
dividing the reported concentration by the lipid weight of the species. In
many cases biota concentrations were already reported on a lipid basis. Units
were equilibrium lipid (ELP) concentrations in g/m3.
2.5. Evaluation of model sensitivity to input parameters
The influence of input parameters on model predictions was assessed by
sensitivity analysis. A total of 165 input parameters were randomly varied
within �10% using a Monte Carlo analysis technique. A uniform distribu-
tion was selected for the sampling of values within this range. A uniform
distribution ensures that all values are equally likely to occur within the
range and the Monte Carlo analysis technique randomly selects values
within the range without applying any weighting factors. The lipid equiva-
lent concentrations of PCB congeners for each food web species were se-
lected as the output to be monitored. One thousand simulation trials were
run using the Crystal Ball� software package for Microsoft Excel� (Crystal
Ball, 2002).
3. Results and discussion
3.1. Comparison between model predictions andmonitoring data
The model was evaluated by comparing model predictedELP concentrations to measured ELP concentrations froma previously developed database for the Baltic (Fig. 1). The
model predictions showed a general increase in ELP concen-trations for all PCB congeners from the base of the foodweb to organisms at higher trophic positions, with the highestlevels predicted in cod and salmon. A difference in bioaccu-mulation behaviour of individual congeners was observedwith PCB 138 and PCB 153 emerging as the most biomagni-fied congeners.
A quantitative model evaluation was performed by calculat-ing the model bias (Eqs. (6) and (7)) as described in Gobas andWilcockson (2003) and Arnot and Gobas (2004).
MBj ¼ 10
�Pn
i¼1
½logðPELP;i=MELP;iÞ�n
�ð6Þ
0.0001
0.001
0.01
0.1
1
phyto
plank
ton
zoop
lankto
n
pelag
ic mac
rofau
na
benth
ic mac
rofau
na
juven
ile he
rring
juven
ile co
d
adult
herri
ng
adult
cod
EL
P,g/
m3
A
0.0001
0.001
0.01
0.1
1
phyto
plank
ton
zoop
lankto
n
pelag
ic mac
rofau
na
benth
ic mac
rofau
na
juven
ile he
rring
juven
ile co
d
adult
herri
ng
adult
cod
EL
P,g/
m3
B
0.0001
0.001
0.01
0.1
1
phyto
plank
ton
zoop
lankto
n
pelag
ic mac
rofau
na
benth
ic mac
rofau
na
juven
ile he
rring
juven
ile co
d
adult
herri
ng
adult
cod
EL
P,g/
m3
C
predicted measured
Fig. 1. (AeC) Comparison between predicted and measured equilibrium lipid
(ELP) concentrations: (A) PCB 28; (B) PCB 101; and (C) PCB 180. Species
are listed in order of estimated trophic position. The error bars represent the
min/max range of measured concentrations.
78 E. Nfon, I.T. Cousins / Environmental Pollution 148 (2007) 73e82
MB¼ 10
264Pm
j¼1
�Pn
i¼1½logðPELPi; j=MELPi; j Þ�
�nm
375
ð7Þ
where MBj is the combined model bias for all PCB congenersin a selected species j, MB is the overall model bias of all PCBcongeners in all species, PELP is the predicted ELP concen-trations (g/m3), and MELP is the measured ELP concentrations(g/m3), i refers to each PCB congener for which PELP andMELP were available, n is the number of chemicals (n¼ 5),m is the number of species included in the model evaluation(m¼ 5; phytoplankton, mizozooplankton, adult sprat, adultcod and salmon). The MB is a measure of the systematicover-prediction (MB> 1) or under prediction (MB< 1) ofthe model. The results of the analysis are presented in Table5. The data presented in Table 5 indicate that the modelover predicted phytoplankton concentrations with an averageMBj of 1.8 for the five selected congeners, although modelpredictions exceeded measured values by a factor of approxi-mately one for PCB 28, PCB 138 and PCB 153 and a factor oftwo for PCB 101 and PCB 180. Predicted concentrations formezozooplankton were close to measured values with anMBj¼ 0.22, an under prediction of a factor of less than two.For the fish species, predictions for adult sprat and salmonwere within a factor of less than two of measured values(MBj 0.03 and 0.12, respectively). Predictions for adult codwere less accurate with an average MBj of 0.73 representinga difference of a factor of five between predicted and mea-sured concentrations. The results for individual congenersshowed an over-prediction of a factor of two and three forPCB 28 and PCB 101, respectively, and a factor of one forPCB 138, PCB 153 and PCB 180. Generally, the evaluationshowed overestimations and underestimations of measuredlipid equivalent concentrations within reasonable limits. Theoverall estimated model bias (MB) for five species and fivePCB congeners was 0.21 indicating an average model under-estimation of less than a factor of two. The 95% confidenceintervals of MB (0.06 and 5.4) express the variability in the ac-curacy of the model predictions among the PCB congeners. Itcan thus be concluded that 95% of the observed congener spe-cific data lie between PELP/18 and PELP� 6.0.
3.2. Role of uptake and loss processes
The relationship between concentration and trophic levelhas previously been shown to be complicated by the differentwater and sediment fugacities. For example, in the model de-veloped by Campfens and Mackay (1997), benthic organismstended to be relatively more contaminated than pelagic organ-ism of a similar trophic status. This is because benthic organ-isms in the Campfens and Mackay (1997) model wereassumed to dwell in the sediment and respire 100% sedimentpore water, which has a higher fugacity than water. In themodel developed in this study, the sediment fugacity (i.e. sed-iment pore water respiration) does not have such a large effecton the contamination of so called ‘‘benthic’’ organisms; ratheruptake by respiration in water dominated the total uptake byrespiration in benthic meiofauna and benthic macrofauna forall PCB congeners. This is a result of the assumed fractionof water respired in the water column (90%) compared tothe fraction of sediment pore water respired (10%).
For the fish species, food was the dominant uptake path-way. A decrease in the percentage contribution of uptakeby respiration with increasing log KOW was observed forthe same species and with trophic levels. For example, up-take by respiration in sprat was 84% of the total uptake ofPCB 28 (log KOW of 5.7), 15% of the total uptake of PCB138 (log KOW of 6.8) and 17% of the total uptake of PCB180 (log KOW of 7.3). Uptake by respiration for cod andsalmon at higher trophic levels were 33 and 22%, respec-tively, PCB 28, 3% and less than 2%, respectively, PCB180. The net chemical uptake through the diet expressed asa fraction of the total uptake was on the average between1 and 8% for the benthic invertebrates and between 60 and99% for the fish species. Dietary uptake was more significantfor the heavier congeners than for the lighter congeners. Forexample, dietary uptake contributed more than 93% of thetotal uptake of the heavier congeners in herring, 95% incod and 99% in salmon.
Loss of contaminants by respiration was the dominant lossprocess at lower trophic levels contributing more than 99% ofthe total loss of contaminants in phytoplankton and zooplank-ton. For the fish species at intermediate trophic positions, themost significant loss processes was loss by excretion while at
Table 5
Results from quantitative model evaluation
log (PELP/MELP)
PCB 28 PCB 101 PCB 138 PCB 153 PCB 180 SDa MBjb CIc
Phytoplankton �0.65 0.34 0.36 �0.05 0.18 0.42 1.09 0.36
Mezozooplankton �1.25 �0.44 �0.53 �0.95 �1.09 0.35 0.14 0.31
Adult sprat �1.66 �1.30 �1.34 �1.89 �1.66 0.25 0.03 0.22
Adult cod 0.48 0.15 �0.45 �0.65 �0.20 0.45 0.73 0.40
Salmon �0.61 �0.65 �0.97 �1.37 �0.97 0.31 0.12 0.27
Overall model bias (MB)d¼ 0.21.a Standard deviation of the ratio of predicted and measured lipid equivalent concentrations.b Model bias for species j combining the results of all PCBs in species j.c The 95% confidence intervals of the geometric mean of the ratio of the predicted and measured lipid equivalent concentrations.d Overall model bias combining the results of all PCBs in five species included in the model performance evaluation.
79E. Nfon, I.T. Cousins / Environmental Pollution 148 (2007) 73e82
higher trophic levels, growth dilution was the dominant lossprocess. Metabolism was insignificant since we assumed an in-finite half-life for PCB congeners in all food web species.
As a means of identifying the abiotic media making themost significant contribution to the final fugacity in the foodweb species, we calculated the contribution of PCB levels inwater and sediment to the total fugacity in food web speciesusing the approach in Sharpe and Mackay (2000). Waterwas the most significant contributor to the total fugacity ofall the species in the food web for all five congeners. PCBlevels in sediment were only significant to the total fugacityin the benthic species and the fish species that interactedwith the benthic species. The result for PCB 180 is shown inFig. 2, which also provides a way of identifying the contribu-tion of direct exposure (by respiration) and indirect exposure(through food web interactions) to the total fugacity of foodweb species. It can be observed that 90% of the fugacity ofthe benthic species is due to PCB levels in water and about10% is due to PCB levels in sediments. For herring, 95% ofthe total fugacity is due to direct uptake by respiration in waterand 5% is obtained indirectly from sediment through interac-tion with benthic species. In cod, between 90 and 92% of thetotal fugacity is due to direct effects of respiration in waterwhile between 8 and 10% of the total fugacity in cod is fromsediment. Finally, 99% of the total fugacity in salmon is dueto respiration in water and 1% is due to PCB 180 in sediments.
3.3. Key input parameters identified by sensitivityanalysis
Only parameters contributing more than 5% to the variancein the predicted concentrations for each species in the foodweb are included in Table 6. The species are listed accordingto trophic levels; each sensitive parameter is indicatedfollowed by the percentage contribution to the variance in
phyt zo
oP.M
aB.M
eB.M
a
J.spr
atJ.H
eJ.C
od
A.herri
ng
A.Cod
salm
on
cont
ribu
tion
to f
ugac
ity
water sediment
Fig. 2. Contribution of PCB 180 levels in abiotic media to total fugacity in
food web species. Species are listed in order of estimated trophic position.
Abbreviations used on the x-axis are as follows. Phyt, phytoplankton; zoo,
zooplankton; P.Ma, pelagic macrofauna; B.Me, benthic meiofauna; B.Ma, ben-
thic macrofauna; J.sprat, juvenile sprat; J.Cod, juvenile cod; A.sprat, adult
sprat; A.He, adult herring; A.Cod, adult cod; Sa, salmon.
the predicted concentration in parentheses, with the mostsensitive first. The percentages indicated are the averages forall five PCB congeners in the model simulation.
Generally, the log KOW and the annual average temperaturedominated the sensitivity analyses with log KOW emerging asthe single most important parameter. Other important parame-ters were the fractional respiration from water and thefractional respiration from sediment pore water. Feeding rate,growth rate and lipid fraction of fish became important forhigher trophic levels (i.e. cod and salmon), but contributedonly about 5e10% of the variance in predicted concentrations.Furthermore, parameters related to the prey of the species athigher trophic levels together contributed between 5 and 80%of the sensitivity data in the fish species. For example, the frac-tional respiration of benthic organisms, sprat and zooplanktonmade significant contributions to the variance in predicted con-centrations in herring, cod and salmon. It is noteworthy that thefractional respiration in pore water of benthic species andsediment related input parameters (organic carbon fraction ofsediment particulates) were particularly important for herringand cod that feed on benthic invertebrates.
It is important that values of the most sensitive inputparameters are of the highest quality in order to limit model un-certainty. The estimated KOW for different PCBs vary widely,depending on the method of estimation. A variation of overa range of about 0.3 log units or more between measured andestimated log KOW values of PCB congeners has been reported(Gusten et al., 1991). As previously discussed, the most reliable
Table 6
Key parameters from sensitivity analysis
Trophic level Species Significant parameters
1 Phytoplankton log KOW (50e70), Temp
(30e40)
Bacteria log KOW (50e70), Temp
(20e40)
2 Zooplankton xW (45e70), log KOW
(15e65), Temp (10e30)
Pelagic macrofauna log KOW (40e60), Temp
(5e40), xW (15e50)
2 and 3 Benthic macrofauna xW (10e85), log KOW
(10e60), Temp (5e30)Benthic meiofauna
3 Sprat xW (5e70), log KOW
(20e60), and Temp (10e30)
3 and 4 Herring log KOW (15e65), Temp
(5e30), xW of He (5e45),
xS of B.Ma (5e35)
4 Cod xW of Me.Z (20e70),
log KOW (5e45), Temp
(5e20), FR (10) GR (5)
Salmon xW of Me.Z (25e80),
log KOW (5e40), Temp
(5e15), xW of sprat (5e25),
FR (10) GR (5), LF (5)
xW is the fractional respiration from water, xS is the fractional respiration from
pore water, FR is the feeding rate, GR and LF is the lipid fraction. All other
parameters are defined in the text. The cut-off criterion for inclusion was
that the parameter contributed at least 5% to the output variance.
80 E. Nfon, I.T. Cousins / Environmental Pollution 148 (2007) 73e82
KOW values were selected from the best sources currentlyavailable so model error due to using unreliable KOW hasbeen limited as much as possible. The high sensitivity of tem-perature is related to its effect on physicalechemical proper-ties. Changes in temperature alter the values of the Henry’slaw constant and KOW. Since Baltic water temperatures arewell known, model error should be low. The largest model errormay well be associated with the fractional respiration fromwater and the fractional respiration from sediment pore water,since as previously discussed, the values used for model inputswere uncertain.
The concentration in water, although only identified to con-tribute less that 5% of the variance in predicted concentrations,is worthy of some further discussion because estimating thisvalue presents special difficulties (Konat and Kowalewska,2001). PCB concentrations in the sub- and low pg/dm3 rangehave been reported for the Baltic (Schulz-Bull et al., 1995;Sobek et al., 2002, 2004). Quantifying such low level concen-trations with certainty is challenging. Secondly, the linearrelationships used in the present model as well as in otherfood web models to describe partitioning, bioconcentrationand bioaccumulation are only valid if the truly dissolvedPCB concentrations are used. Using total concentrations orconcentrations determined in the presence of colloids or dis-solved organic matter could introduce non-linearity in the bio-accumulation relationships. For example, Borga et al. (2005)attributed the variability in predicted BAFs in copepods col-lected from the North American great Lakes to analytical con-straints in measuring very low concentrations of contaminantsin the water column. Analytical methodology has been im-proving over the years, for example stainless steel equipmenthas been replaced by HDPE or silicon tubing to limit sorptionto sampling equipment, and sample handling procedures arecurrently being improved (Sobek, 2005). In spite of these im-provements, the extensiveness of the analytical process im-plies that errors might occur at different stages and it istherefore possible that our input water concentrations area source of model error.
The present model also assumes that sorption is principallyby dissolution to the organic carbon fraction of sediments andparticulate matter. However, it has been suggested that suchequilibrium partitioning approaches are too simplistic becausethey do not account for the slow sorption/desorption kineticsof some organic compounds (Pignatello and Xing, 1996).Furthermore, organic carbon may not be the only part of a ben-thic or suspended particulate that can strongly sorb organiccompounds. It has emerged over the last decade that soot car-bon, carbonaceous organic materials, black-carbon, coal andkerogen may have a stronger sorption capacity than organiccarbon and may be of significance in the adsorption of hydro-phobic organic chemicals in the environment (Barring et al.,2002; Cornelissen et al., 2005). Equations have been derivedfor estimating the sorption of organic compounds to sootcarbon and to date has only been validated experimentallyfor the polycyclic aromatic hydrocarbons and are not yetwidely applicable to other organic chemicals (Qiu and Davis,2004; Cornelissen et al., 2005). In this study we assumed that
soot sorption is not important for dilute marine systems(Cornelissen and Gustafsson, 2004), such as the open Baltic,and that equilibrium sorption to particulate organic carbon isan appropriate simplification.
4. Concluding remarks
The modelling exercise presented here provided a mecha-nistic and quantitative understanding of the uptake and lossprocesses affecting PCB levels in a large Baltic food web.Key model input parameters for different species were alsoidentified from the sensitivity analysis. It should be noted,however, that the model applies simplifying assumptions andtherefore has limitations. For example, the steady state as-sumption in the current model version does not account forthe effect of temporal differences in contaminant levels (andthus exposure) due to changing chemical emissions. Anothersimplifying assumption was that the model input exposureconcentrations for water and sediment were averages for theentire Baltic. A possible model improvement could considerthe division of the Baltic into different basins as has beendone in the POPCYCLING-Baltic Model (Wania et al.,2000). This would allow different water and sediment concen-trations to be defined for each basin. A disadvantage of thisapproach is that it would no longer be possible to poolbiomonitoring data from different Baltic basins for modelevaluation.
The results of the model sensitivity analysis provide a guideto model improvements. Little can be currently done toimprove the input values of the two most sensitive model inputparameters, namely KOW and temperature, since best availablevalues were used. We suggest that the focus of future effortsshould be on better describing the fractional respiration fromwater and pore water, which, as discussed, is currently a highlyuncertain input parameter.
Finally, since the model is mechanistic, it can potentially beused to estimate the food web bioaccumulation of a widerange of other non-ionic organic contaminants that are presentin the Baltic Sea. We therefore recommend further model eval-uation using other compound classes as more data becomeavailable. The model could be a useful tool for estimatingwildlife exposure to organic contaminants as well as humanexposure from consumption of fish.
Acknowledgements
This study was financially supported by FORMAS (theSwedish Research Council for Environment, AgriculturalSciences and Spatial Planning) through grant number 21.0/2003-0206. The authors thank Costas Prevedouros forproviding valuable comments and suggestions.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.envpol.2006.11.033.
81E. Nfon, I.T. Cousins / Environmental Pollution 148 (2007) 73e82
References
Arnot, J.A., Gobas, F.A.P.C., 2003. A generic QSAR for assessing the bioac-
cumulation potential of organic chemicals in aquatic food webs. QSAR
Comb. Sci. 22, 337e345.
Arnot, J.A., Gobas, F.A.P.C., 2004. A food web bioaccumulation model for organic
chemicals in aquatic ecosystems. Environ. Toxicol. Chem. 23, 2343e2355.
Beyer, A., Wania, F., Gouin, T., Mackay, D., Matthies, M., 2002. Selecting
internally consistent physicochemical properties of organic compounds.
Environ. Toxicol. Chem. 21, 941e953.
Bignert, A., Olsson, M., Persson, W., Jensen, S., Zakrisson, S., Litzen, K.,
Eriksson, U., Haggberg, L., Alsberg, T., 1998. Temporal trends of organ-
ochlorines in Northern Europe, 1967e1995. Relation to global fraction-
ation, leakage from sediments and international measures. Environ.
Pollut. 99, 177e198.
Borga, K., Fisk, A.T., Hargrave, B., Hoekstra, P.F., Muir, D.C.G., 2004.
Biological and chemical factors of importance in the bioaccumulation
and trophic transfer of persistent organochlorine contaminants in Arctic
marine food webs. Environ. Toxicol. Chem. 23, 2367e2385.
Borga, K., Fisk, A.T., Hargrave, B., Hoekstra, P.F., Swackhamer, D.,
Muir, D.C.G., 2005. Bioaccumulation factors for PCBs revisited. Environ.
Sci. Technol. 39, 4523e4532.
Breitholtz, M., Hill, C., Bengtsson, B.E., 2001. Toxic substances and reproduc-
tive disorders in Baltic fish and crustaceans. Ambio 4, 210e216.
Broman, D., Naf, C., Rolff, C., Zebuhr, Y., Fry, B., Hobbie, J., 1992. Using
ratios of stable nitrogen isotopes to estimate bioaccumulation and flux of
polychlorinated dibenzo-p-dioxins (PCDDs) and dibenzofurans (PCDFs)
in two food chains from the northern Baltic. Environ. Toxicol. Chem.
11, 331e345.
Burreau, S., Zebuhr, Y., Broman, D., Ishaq, R., 2004. Biomagnification of
polychlorinated biphenyls (PCBs) and polybrominated diphenyl ethers
(PBDEs) studied in pike (Esox lucius), perch (Perca fluviatilis) and roach
(Rutilus rutilus) from the Baltic Sea. Chemosphere 55, 1043e1052.
Barring, H., Bucheli, T.D., Broman, D., Gustafsson, O., 2002. Soot-water dis-
tribution coefficients for polychlorinated dibenzo-p-dioxins, polychlori-
nated dibenzofurans and polybrominated diphenylethers determined with
the soot co solvency-column method. Chemosphere 49, 515e523.
Campfens, J., Mackay, D., 1997. Fugacity based model of PCB bioaccumula-
tion in complex aquatic food webs. Environ. Sci. Technol. 31, 577e583.
Clark, K.E., Gobas, F.A.P.C., Mackay, D., 1990. Model of organic chemical
uptake and clearance by fish from food and water. Environ. Sci. Technol.
24, 1203e1213.
Clark, K.E., Mackay, D., 1991. Dietary uptake and biomagnification of four
chlorinated hydrocarbons by guppies. Environ. Toxicol. Chem. 10,
1205e1217.
Cornelissen, G., Gustafsson, O., 2004. Sorption of phenanthrene to environ-
mental black carbon in sediment with and without organic matter and
native sorbates. Environ. Sci. Technol. 38, 148e155.
Cornelissen, G., Gustafsson, O., Bucheli, T.D., Jonker, M.T.O.,
Koelmans, A.A., Van Noort, P.C.M., 2005. Extensive sorption of organic
compounds to black carbon, coal, and kerogen in sediments and soils:
mechanisms and consequences for distribution, bioaccumulation, and bio-
degradation. Environ. Sci. Technol. 39, 6881e6895.
Crystal Ball� Standard Edition 2002.2. Decisionering, Inc., Denver, Colorado,
USA.
Czub, G., Mclachlan, M.S., 2004. A food chain model to predict the levels of
lipophilic organic contaminants in humans. Environ. Toxicol. Chem. 23,
2356e2366.
Ejdung, G., 1998. Predatory Processes in Baltic Benthos. Doctoral Disserta-
tion. Department of Zoology, Stockholm University, Sweden. ISBN: 91-
87272-69-5.
Endicott, D., Kreis, R.G., Mackelburg, L., Kandt, D., 1998. Modelling PCB
bioaccumulation by the zebra mussel (Dreissena polymorpha) in Saginaw
Bay, Lake Huron. J. Great Lakes Res. 24, 411e426.
Fraser, A.J., Burkow, I.C., Wolkers, H., Mackay, D., 2002. Modelling the
biomagnification and metabolism of contaminants in harp seals of the
Barents Sea. Environ. Toxicol. Chem. 21, 55e61.
Gobas, F.A.P.C., Zhang, X., Wells, R., 1993. Gastrointestinal magnification:
the mechanism of biomagnification and food chain accumulation of
organic chemicals. Environ. Sci. Technol. 27, 2855e2863.
Gobas, F., Wilcockson, J., 2003. San Francisco Bay PCB Food Web Model.
RMP Technical Report. SFEI Contribution 90 December 2003.
Gobas, F.A.P.C., 1993. A model for predicting the bioaccumulation of hydro-
phobic organic chemicals in aquatic food-webs: application to Lake On-
tario. Ecol. Modell. 69, 1e17.
Gorokhova, E., Fagerberg, T., Hansson, S., 2004. Predation by herring (Clupea
harengus) and sprat (Sprattus sprattus) on Cercopagis pengoi in a western
Baltic Sea bay. ICES J. Mar. Sci. 61, 959e965.
Granskog, M.A., 1999. Baltic Sea ice as a medium for storage of particulate
matter and elements. ICES J. Mar. Sci. 56, 172e175.
Gschwend, P.M., Wu, S., 1985. On the constancy of sediment-water partition
coefficients of hydrophobic organic pollutants. Environ. Sci. Technol. 19,
90e96.
Gusten, H., Horvatic, D., Sabljic, A., 1991. Modelling n-octanol/water parti-
tion coefficients by molecular topology: polycyclic aromatic hydrocarbons
and their alkyl derivatives. Chemosphere 23, 199e213.
Hakanson, L., Gyllenhammar, A., Brolin, A., 2004. A dynamic compartment
model to predict sedimentation and suspended particulate matter in coastal
areas. Ecol. Modell. 175, 353e384.
Hansson, S., Karlsson, L., Ikonen, E., Christiansen, O., Mitans, A., Uzars, D.,
Petersson, E., Ragnarsson, B., 2001. Stomach analysis from Baltic salmon
from 1959e1962 and 1994e1997: possible relations between the diet and
yolk-sac-fry mortality (M174). J. Fish Biol. 58, 1730e1745.
Harvey, C.J., Cox, S.P., Essington, T.E., Hansson, S., Kitchell, J.F., 2003.
An ecosystem model of food web and fisheries interactions in the Baltic
Sea. ICES J. Mar. Sci. 60, 939e950.
Hill, C., 1991. Mechanisms Influencing the Growth, Reproduction and Mortal-
ity of Two Co-occurring Amphipod Species in the Baltic Sea. Ph.D. thesis,
Department of Zoology, Stockholm University, Sweden. ISBN: 91-87272-
27-X.
Huttig, J., Oehme, M., 2005. Presence of chlorinated paraffins in sediments
from the North and Baltic Seas. Arch. Environ. Contam. Toxicol. 49,
449e456.
ICES, 2001. Report of the Baltic Fisheries Assessment Working group. ICES
CM 2001/ACFM: 18 ICES Advisory Committee on Fishery
Management.
Jonsson, A., Carman, R., 2000. Distribution of PCBs in sediment from differ-
ent bottom types and water depths in Stockholm archipelago, Baltic Sea.
Ambio 29, 277e281.
Kelly, B.C., Gobas, F.A.P.C., McLachlan, M.S., 2004. Intestinal absorption
and biomagnification of organic contaminants in fish, wildlife, and hu-
mans. Environ. Toxicol. Chem. 23, 2324e2336.
Kiviranta, H., Vartiainen, T., Parmanne, R., Hallikainen, A., Koistinen, J.,
2003. PCDD/Fs and PCBs in Baltic herring during the 1990s. Chemo-
sphere 50, 1201e1216.
Kjeller, L.O., Rappe, C., 1995. Time trends in levels, patterns and profiles
for polychlorinated dibenzo-p-dioxins, dibenzofurans, and biphenyls in
a sediment core from the Baltic proper. Environ. Sci. Technol. 29,
346e355.
Koistinen, J., Koivusaari, J., Nuuja, I., Paasivirta, J., 1995. PCDEs, PCBs,
PCDDs and PCDFs in black guillemots and white tailed sea eagles from
the Baltic Sea. Chemosphere 30, 1671e1684.
Konat, J., Kowalewska, G., 2001. Polychlorinated biphenyls (PCBs) in
sediments of the Southern Baltic Sea e trends and fate. Sci. Total Environ.
280, 1e15.
Li, N., Wania, F., Lei, Y.D., Daly, G.L., 2003. A comprehensive and critical
compilation, evaluation and selection of physicalechemical property
data for selected polychlorinated biphenyls. J. Phys. Chem. Ref. Data
32, 1545e1590.
Mackay, D., 2001. Multimedia Environmental Models: The Fugacity
Approach, second ed. Lewis Publishers, Boca Raton, London, New
York, Washington DC.
Mackay, D., Fraser, A., 2000. Bioaccumulation of persistent organic chemi-
cals: mechanisms and models. Environ. Pollut. 110, 375e391.
82 E. Nfon, I.T. Cousins / Environmental Pollution 148 (2007) 73e82
Mackay, D., Shiu, W.Y., Ma, K.C., 1992. Illustrated Handbook of PhysicaleChemical Properties and Environmental Fate for Organic Chemicals.
Volume 1: Monoaromatic Hydrocarbons, Cholorobenzenes and PCBs.
Lewis Publishers.
Mackintosh, C.E., Maldonado, J., Hongwu, J., Hoover, N., Chong, A.,
Gobas, F.A.P.C., 2004. Distribution of phthalate esters in a marine aquatic
food web: comparison to polychlorinated biphenyls. Environ. Sci. Technol.
38, 2011e2020.
Meylan, W.M., Howard, P.H., Boethling, R.S., Aronson, D., Printup, H.,
Gouchie, S., 1999. Improved method for estimating bioconcentration/bio-
accumulation factor from octanolewater partition coefficient. Environ.
Toxicol. Chem. 18, 664e672.
Morrison, H.A., Gobas, F.A.P.C., Lazar, R., Haffner, G.D., 1996. Development
and verification of a bioaccumulation model for organic contaminants in
benthic invertebrates. Environ. Sci. Technol. 30, 3377e3384.
Morrison, H.A., Gobas, F.A.P.C., Lazar, R., Whittle, D.M., Haffner, G.D.,
1997. Development and verification of a benthic/pelagic food web bioac-
cumulation model for PCB congeners in Western Lake Erie. Environ.
Sci. Technol. 31, 3267e3273.
Neely, B.W., Branson, D.R., Blau, G.E., 1974. Partition coefficients to mea-
sure bioconcentration potential of organic chemicals in fish. Environ.
Sci. Technol. 8, 1113e1115.
Nfon, E., Cousins, I.T., 2006. Interpreting time trends and biomagnification of
PCBs in the Baltic region using the equilibrium lipid partitioning ap-
proach. Environ. Pollut. 144, 994e1000.
Nyman, M., Koistinen, J., Fant, M.L., Vartiainen, T., Helle, E., 2002. Current
levels of DDT, PCB, and trace elements in the Baltic ringed seals (Phocahispida baltica) and grey seals (Halichoerus grypus). Environ. Pollut. 119,
399e412.
Pignatello, J.J., Xing, B., 1996. Mechanisms of slow sorption of organic chem-
icals to natural particles. Environ. Sci. Technol. 30, 1e11.
Qiu, X., Davis, J.W., 2004. Environmental bioavailability of hydrophobic
organochlorines in sediments e a review. Remediat. J. 14, 55e84.
Rodhe, J., Winsor, P., 2002. On the influence of the fresh water supply on the
Baltic Sea mean salinity. Tellus 54A, 175e186.
Rolff, C., Strandberg, B., Zeblihr, Y., Zook, D., Rappe, C., 1995. Levels of
PCBs in the aquatic environment of the Gulf of Bothnia: benthic species
and sediments. Mar. Pollut. Bull. 32, 210e218.
Roots, O., Talvari, A., 1997. Bioaccumulation of toxic chlororganic com-
pounds and their isomers into the organism of Baltic grey seal. Chemo-
sphere 35, 979e985.
Schwarzenbach, R.E., Gschwend, P.M., Imboden, D.M., 2003. Environmental
Organic Chemistry, second ed. John Wiley & Sons, Inc., New York.
Sharpe, S., Mackay, D., 2000. A framework for evaluating bioaccumulation in
food webs. Environ. Sci. Technol. 34, 2373e2379.
Sinkkonen, S., Paasivirta, J., 2000. Degradation half-life times of PCDDs,
PCDFs and PCBs for environmental fate modelling. Chemosphere 40,
943e949.
Schulz-Bull, D.E., Petrick, G., Kannan, N., Duinke, J.C., 1995. Distribution of
individual chlorobiphenyls (PCB) in solution and suspension in the Baltic
Sea. Mar. Chem. 48, 245e270.
Snelgrove, P.V.R., 1999. Getting to the bottom of marine biodiversity: sedi-
mentary habitats. BioScience 49, 129e138.
Sobek, A., Gustafsson, O., Axelman, J., 2002. An evaluation of the importance
of the sampling step to the total analytical variance e a four system field
based sampling intercomparison study for hydrophobic organic contami-
nants in the surface waters of the open Baltic Sea. Int. J. Environ. Anal.
Chem. 83, 177e187.
Sobek, A., Gustafsson, O., Hajdu, S., Larsson, U., 2004. Particleewater parti-
tioning of PCBs in the photic zone: a 25 month study in the open Baltic
Sea. Environ. Sci. Technol. 38, 1375e1382.
Sobek, A., 2005. Uptake Processes of Polychlorinated Biphenyls at the Base of
the Pelagic Food Web. Ph.D. thesis, Stockholm University, Sweden. ISBN:
91-7155-040-2.
Stevenson, R.W., 2003. Development and Application of a Model Describing
the Bioaccumulation and Metabolism of Polycyclic Aromatic Hydrocar-
bons in Marine Benthic Food Web. A project submitted in partial fulfill-
ment for the degree of Master of Resource Management. Report No.
334. Simon Fraser University.
Strandberg, B., Bandh, C., Van Bavel, B., Bergqvist, P.A., Broman, D.,
Naf, C., Pettersen, H., Rappe, C., 1998. Concentrations, biomagnifica-
tion and spatial variation of organochlorine compounds in a pelagic
food web in the northern part of the Baltic Sea. Sci. Total Environ.
217, 143e154.
Thomann, R.V., Connolly, J.P., 1984. Model of PCB in the Lake Michigan lake
trout food chain. Environ. Sci. Technol. 18, 65e71.
Thomann, R.V., Conolly, J.P., Pakerton, T.F., 1992. An equilibrium model for
organic chemical accumulation in aquatic food webs with sediment inter-
action. Environ. Toxicol. Chem. 11, 257e267.
Wania, F., Persson, J., Di Guardo, A., McLachlan, M.S., 2000. The POPCY-
CLING-Baltic Model. A non-steady state multi-compartment mass balance
model of the fate of persistent organic pollutants in the Baltic Sea
environment.
Winsor, M.H., Boese, B.L., Lee, H., Randall, R.C., Specht, D.T., 1990. Deter-
mination of the ventilation rates of interstitial and overlying water by the
clam Macoma nasuta. Environ. Toxicol. Chem. 9, 209e213.