Continental scale passive air sampling of persistent organic pollutants using rapidly equilibrating thin films (POGs) Nick J. Farrar a , Konstantinos Prevedouros a,1 , Tom Harner b , Andrew J. Sweetman a , Kevin C. Jones a, * a Department of Environmental Science, Institute of Environmental and Natural Sciences, Lancaster University, Lancaster, LA1 4YQ, UK b Science & Technology Branch, Environment Canada, 4905 Dufferin Street, Toronto, ON M3H 5T4, Canada Received 22 September 2005; accepted 20 December 2005 Polymer-coated glass (POG) slides are used as short-term air samplers in a survey of POPs in different European countries. Abstract A novel design of rapidly equilibrating passive air sampler was deployed at 38 sites across 19 European countries to investigate short-term spatial variability of persistent organic pollutants (POPs). Devices were sealed in airtight containers to eliminate the possibility of contamination during transit and couriered to recipients with deployment instructions. Exposure times of 7 days permitted the use of back trajectory analysis to further understand the factors responsible for influencing the large-scale spatial distribution of PCBs, PBDEs, PCNs, PAHs, lindane and HCB. Following sampler harvest, devices were sealed and returned for analysis. Comparison of sequestered levels showed that PAHs exhibited the greatest spatial variability (by a factor of 30) with higher levels often associated with greater population density. In contrast, HCB values were much more uniform, reflecting its well mixed distribution in the atmosphere. Spatial variation was strongly influenced by air mass origin, with lower levels being observed at most sites impacted by maritime air masses. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Passive air sampling; Persistent organic pollutants; Polymer-coated glass; Continental scale 1. Introduction In an effort to further understand the behaviour of persistent organic pollutants (POPs), researchers have focused their ener- gies on trying to unravel not only their sources and fates within the environment, but also the interactions of these com- pounds with environmental media (Cousins et al., 1999; Bar- ber et al., 2003; Meijer et al., 2003a). Research into these areas is largely driven by international concern over the toxic- ity and long-range atmospheric transport (LRAT) potential of both ‘‘new’’ and ‘‘old’’ POPs, and the consequential need to identify the processes responsible for global distribution and food chain exposure. It is vital to assess the distribution of POPs when attempt- ing to investigate any potential source/sink relationships that may exist within a given region. If such relationships become established, methods of interaction between compound and environment can be inferred. Workers have attempted to do this on a variety of scales using computer based models to sim- ulate the atmospheric mixing of pollutants from point sources and hence determine local and regional scale contaminant dis- tribution (MacLeod and Mackay, 2004; Prevedouros et al., 2004). However, such techniques are subject to large uncer- tainties (ApSimon et al., 2000; Fenner et al., 2004) unless the model can be validated using measured air concentration data. This requires sampling being carried out simultaneously at numerous locations, using a reproducible and calibrated * Corresponding author. Tel.: þ44 1524 593972; fax: þ44 1524 593985. E-mail address: [email protected](K.C. Jones). 1 Present address: Department of Applied Environmental Science (ITM), Stockholm University, SE-10691 Stockholm, Sweden. 0269-7491/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.envpol.2005.12.057 Environmental Pollution 144 (2006) 423e433 www.elsevier.com/locate/envpol
11
Embed
Continental scale passive air sampling of persistent organic pollutants using rapidly equilibrating thin films (POGs)
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Continental scale passive air sampling of persistent organic pollutantsusing rapidly equilibrating thin films (POGs)
Nick J. Farrar a, Konstantinos Prevedouros a,1, Tom Harner b,Andrew J. Sweetman a, Kevin C. Jones a,*
a Department of Environmental Science, Institute of Environmental and Natural Sciences, Lancaster University, Lancaster, LA1 4YQ, UKb Science & Technology Branch, Environment Canada, 4905 Dufferin Street, Toronto, ON M3H 5T4, Canada
Received 22 September 2005; accepted 20 December 2005
Polymer-coated glass (POG) slides are used as short-term air samplers in a survey of POPsin different European countries.
Abstract
A novel design of rapidly equilibrating passive air sampler was deployed at 38 sites across 19 European countries to investigate short-termspatial variability of persistent organic pollutants (POPs). Devices were sealed in airtight containers to eliminate the possibility of contaminationduring transit and couriered to recipients with deployment instructions. Exposure times of 7 days permitted the use of back trajectory analysis tofurther understand the factors responsible for influencing the large-scale spatial distribution of PCBs, PBDEs, PCNs, PAHs, lindane and HCB.Following sampler harvest, devices were sealed and returned for analysis. Comparison of sequestered levels showed that PAHs exhibited thegreatest spatial variability (by a factor of 30) with higher levels often associated with greater population density. In contrast, HCB valueswere much more uniform, reflecting its well mixed distribution in the atmosphere. Spatial variation was strongly influenced by air mass origin,with lower levels being observed at most sites impacted by maritime air masses.� 2006 Elsevier Ltd. All rights reserved.
In an effort to further understand the behaviour of persistentorganic pollutants (POPs), researchers have focused their ener-gies on trying to unravel not only their sources and fateswithin the environment, but also the interactions of these com-pounds with environmental media (Cousins et al., 1999; Bar-ber et al., 2003; Meijer et al., 2003a). Research into theseareas is largely driven by international concern over the toxic-ity and long-range atmospheric transport (LRAT) potential ofboth ‘‘new’’ and ‘‘old’’ POPs, and the consequential need to
E-mail address: [email protected] (K.C. Jones).1 Present address: Department of Applied Environmental Science (ITM),
Stockholm University, SE-10691 Stockholm, Sweden.
0269-7491/$ - see front matter � 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.envpol.2005.12.057
identify the processes responsible for global distribution andfood chain exposure.
It is vital to assess the distribution of POPs when attempt-ing to investigate any potential source/sink relationships thatmay exist within a given region. If such relationships becomeestablished, methods of interaction between compound andenvironment can be inferred. Workers have attempted to dothis on a variety of scales using computer based models to sim-ulate the atmospheric mixing of pollutants from point sourcesand hence determine local and regional scale contaminant dis-tribution (MacLeod and Mackay, 2004; Prevedouros et al.,2004). However, such techniques are subject to large uncer-tainties (ApSimon et al., 2000; Fenner et al., 2004) unlessthe model can be validated using measured air concentrationdata. This requires sampling being carried out simultaneouslyat numerous locations, using a reproducible and calibrated
technique to create maps of chemical distribution, which givea ‘‘snapshot’’ of chemical distribution and spatial variability.Although widely used and accurate, the large numbers of con-ventional high volume (hi-vol) air samplers needed to do thiswould be prohibitively expensive and logistically challenging,as each unit requires both power and skilled technicians to en-sure smooth operation (Shoeib and Harner, 2002).
Environmental media have been used in the past to producesuch maps of concentration distribution (Calamari et al., 1991;Tremolada et al., 1996; Holoubek et al., 2000). However,inconsistency in contaminant uptake rates, high biological var-iability and variable exposure times may all make interpreta-tion very difficult. The ability to produce a relativelyuniform device with similar uptake characteristics wouldtherefore aid in the comparison of data from different loca-tions. The development of passive samplers over the lastdecade has allowed workers to deploy devices in locationspreviously considered too remote for routine air monitoring.Of the various types of sampler available, most (Ockendenet al., 2001; Wania et al., 2003; Jaward et al., 2004) areused to integrate air concentrations over large time scales(monthseyears), which facilitates the investigation ofsource/sink relationships. These ‘‘kinetic’’ devices constitutethe majority of samplers routinely deployed to monitor airconcentrations, as the long equilibration times allow them tobe left unattended for large periods of time. However, fastreacting, ‘‘equilibrium’’ samplers that rapidly adjust to changesin ambient air concentrations are more desirable whenattempting to better understand specific contaminant transportand dispersion events (Harner et al., 2003). Short-term airsampling can be combined with meteorological and air massback trajectory data to give added power to distribution studiesand dispersion modelling.
POGs (POlymer-coated Glass) are a new design of rapidlyequilibrating passive sampler, created by applying a thin poly-meric stationary phase (ethylene vinyl acetatedEVA) toa solid glass structure (Harner et al., 2003). The high ‘‘surfacearea to volume’’ ratio (SA:V) of these samplers, which isgreater than previous designs, results in significantly reducedequilibration times; PCB-28 is seen to reach equilibriumwithin hours, for example, although heavier PCBs take lon-gerdtypically several days (Fig. 1). Unlike other devices,the thickness of the stationary phasedand hence itsSA:Vdcan be chosen prior to deployment, depending onthe objectives of the campaign. By producing a thick coatingof EVA, the increased capacity of the sampler ensures thatthe linear (kinetic) phase of uptake is lengthened. By increas-ing the time to equilibrium, the POG can be used to investigatesource-sink relationships because short-term variation in airconcentrations will become integrated. Conversely, a thinEVA coating would reduce equilibration times, thus allowingthe study of short-term flux processes.
Studies using POGs are still in their infancy. The firstcharacterised its uptake characteristics (Harner et al., 2003),but there has been only one study demonstrating the utilityof the device as a tool for mapping the spatial variabilityof POPs (Farrar et al., 2005a). By positioning the samplers
at different heights throughout the urban boundary layer,the vertical concentration profiles were obtained for a rangeof compounds, allowing inferences to be made regarding theirorigin. Although only deployed on a small scale, this studyclearly showed the sensitivity of the POG and illustratedhow applicable the technique is for spatial mapping. Thestudy reported here takes the deployment one step furtherby sampling POPs at 38 sites in 19 European countriessimultaneously.
Previous studies have attempted to clarify the sources anddistribution of POPs within the environment, but the unique-ness of this campaign is its ability to synchronise the investi-gation of numerous compounds simultaneously. A range offactors undoubtedly affect the spatial distribution of POPson a continental scale; past/current usage, population density,temperature, land use and air mass origin all play a role. A keyquestion for regulators is the degree to which ongoing primarysources control ambient concentrations. This study focuses ona wide range of POPs (PCBs, PAHs, PCNs, PBDEs and OCPs)with different sources (e.g. industrial; combustion; agriculture)and properties (e.g. atmospheric persistence; volatility), tohighlight their contrasting nature and the versatility of POGsampling. In conjunction with a parallel study (Jaward et al.,2004), this paper presents data from the first short-term, wide-spread deployment of rapidly equilibrating thin films acrossEuropedthe Passive Air Sampling Across Europe (PASAE)campaign.
2. Principles of POG sampling
The principles behind passive air sampling for POPs havebeen discussed in some detail elsewhere (Shoeib and Harner,2002). In brief, gas phase compounds partition into the sam-pling medium (in this case, the EVA) during an uptake phaseanddif the exposure time is long enoughdapproach an equi-librium, which is in turn a function of temperature. Becauseuptake is air-side controlled (Harner et al., 2003; Bartkowet al., 2004) variations in wind speed are a potentially con-founding factor. However, samplers can be deployed in cham-bers to buffer the flow of air (in an effort to normalise uptakerates). Furthermore, the total capacity of the sampler at
Am
ount
of
com
poun
dse
ques
tere
d
Time
Linear Phase Equilibrium Phase
Curvi-Linear Phase
Fig. 1. Idealised uptake regime for POPs, showing each phase of uptake.
425N.J. Farrar et al. / Environmental Pollution 144 (2006) 423e433
equilibrium differs between compounds; chemicals withhigher equilibrium partition coefficients will take longer toreach equilibrium. The capacity and time to equilibrium canalso be varied by adjusting the EVA film thicknessdthickerfilms will obviously take longer to reach equilibrium due tothe increased capacity to retain POPs. Recent studies onselected organochlorine pesticides have shown that KEVA-AIR
(the EVA air partition coefficient) and its variation with tem-perature correlates with the octanol:air partition coefficient(KOA) (Wilcockson and Gobas, 2001). Thus KOA values thatare known for several classes of POPs can be used to estimateKEVA-AIR.
Previous studies (Harner et al., 2003) have utilised anexperimentally determined uptake rate of 3 m3/day to backcalculate the air concentration for compounds within the linearphase, whilst for compounds that had equilibrated, back calcu-lating air concentrations relied upon the sequestered mass ofa compound and KEVA-AIR. However, this assumes that thesampling mode for each compound is knowndand thisrequires calibration. Recently, advances in estimating the airconcentration have been made by calculating the effectiveair volume needed to supply enough chemical to the POG toresult in an equilibrium state between the air and the polymer(Harner et al., 2004; Farrar et al., 2005a,b). The volume of airrequired to establish equilibrium between the sampler and theair (VAIR) is calculated using:
where AEVA and VEVA are the planar surface area (cm2) andvolume (cm3) of EVA respectively, kA is the air-side masstransfer coefficient (cm/day) and t is the sampling period(days). This theory relies upon the principle that the volumeof air needed to achieve equilibrium with the POG is relatedto the EVA air partition coefficient and the capacity of thefilm to retain the compound. For example, a PCB witha high KOA (and hence, a high KEVA-AIR) will have a greaterchemical capacity in the EVA film than a compound witha lower partition coefficient, and will therefore requirea greater volume of air to supply sufficient mass of PCB tothe POG. This results in compounds with lower KEVA-AIR
values equilibrating in a shorter period of time. Figs. 1 and2 illustrate the points raised in this section.
3. Materials and methods
3.1. POG manufacture and deployment
POGs were manufactured by following a method detailed by Harner et al.
(2003). A solution of ethylene vinyl acetate (EVA, Elvax 40W, DuPont Can-
ada) was prepared by dissolving 4 g of EVA pellets in 200 ml of analytical
grade dichloromethane (DCM). PCB-40 and PCB-155 were spiked into this
coating solution to act as performance reference compounds (PRCs); both
were chosen primarily because of their low atmospheric abundance. This
solution was transferred to a 500 ml coating vessel, pre-rinsed with DCM, hex-
ane and acetone.
Glass cylinders, pre-rinsed with the above three solvents, were dipped into
the coating solution, ensuring that approximately 80% of the POG was cov-
ered. Once removed from the solution, DCM evaporated to leave a layer of
EVA adhering to the glass surface, covering approximately 280 cm2. POGs
were then attached to the sampling chamber and sealed within an airtight
enclosure by covering with a larger jar with a Teflon-lined lid.
In total 38 samplers were prepared, placed in airtight PVC boxes and dis-
patched, by courier, to selected recipients in 19 countries throughout Europe,
with instructions to deploy the samplers from 13 June 2002 to 20 June 2002
(see Fig. 3 for site locations). Shipping POGs before the start date allowed
each participant to expose the samplers simultaneously, but obviously, avoid-
ance of contamination during transport and storage was critical. Each recipient
chose the deployment site, in line with advice about proximity to buildings and
height from the ground. After 1 week, POGs were harvested, sealed and cour-
iered back to Lancaster University for extraction.
3.2. Extraction, clean-up and analysis
POGs were placed in a sealed vessel and repeatedly soaked/rinsed with
250 ml of DCM to ensure that the EVA coating was fully stripped from the
surface of the POG. To this solution, deuterated PAHs (1-methylnaphtha-
L, CL or E indicate that the compound lies in the ‘‘linear’’, ‘‘curvilinear’’ or ‘‘equilibrium’’ phase of uptake after 7 days of exposure. The ‘‘effective air volume’’
(m3) is the air volume required to supply a given compound to reach gas phase: POG equilibrium at 20 �C. The figure in parentheses is the time calculated (in
hours) to reach that equilibrium. Derived air concentrations, determined for the pg/sampler data, are as described in the text. <DL, below detection limit; N/c, not
calculated; N/a, not available.a KOA were taken from Harner and Shoeib, 2002 (PBDEs), Harner and Bidleman, 1996 (PCBs), Harner and Bidleman, 1998 (PAHs and PCNs) and Harner and
Mackay, 1995 (HCB).b Geometric mean (arithmetic mean: range).c Geometric mean and range.
Fig. 4. Back trajectories across Europe, from 13 June 2002 to 20 June 2002.
The data in this form highlights differences between exposedsamplers and the detection limits between compounds.A number of compounds were always in the sampler, includ-ing several PCB congeners, HCB, lighter PAHs, PCNs andPBDE-99 (see % detected in Table 1). Achieving this fromonly 7 days of deployment emphasises the sensitivity ofPOG sampling and analytical techniques for monitoringand detecting POPs from low air volumes.
Although measured air concentrations of POPs have beenshown to vary throughout Europe (Meijer et al., 2003b; Jawardet al., 2004), with high and low levels often associated withurban and rural regions respectively, the average (geometric)concentrations (pg/sampler) for different compounds werequite similar. For example, different PCB congeners coveringa wide range of KOA varied only by a factor of 2e3 whenexpressed as pg/sampler (e.g. levels between 38 pg/sampler(PCB-49) and 135 pg/sampler (PCB-153)). Even thoughPCN concentrations within the UK atmosphere are typicallylower than PCBs by a factor of w5, the levels of PCBs andPCNs on the samplers were similar. Greatest-between com-pound differences were seen for the PAHs, with benzo[a]py-rene (17 pg/sampler) being lower than phenanthrene (730 pg/sampler) by a factor of 43. Phenanthrene is usually the mostabundant PAH in the atmosphere (Lee and Jones, 1999) andexists almost exclusively in the gas phase. In contrast, benzo[a]pyrene is much less abundant in air and exists almostexclusively on particles. The design of the sampling chamberprotects the POG from the sun and rain, yet allows the free-flow of air around the POG surface. This means that onlyvery fine particulates (<100 nm) actually migrate to the sam-pler surface.
SPAHs exhibited substantial variability (920e28100 pg/sampler) (see Fig. 6) whereas HCB (a very volatile OCPwith a low KOA) only varied between 30e170 pg/sampler(i.e. a factor of <6). This relatively even distribution ofHCB throughout Europe (Fig. 5) substantiates findings fromother studies (Jaward et al., 2004). HCB volatilises fromsoil, post application and has high LRAT potential (Wangand Jones, 1994; Bailey, 2001). PAHs, on the other hand,are found in urban ‘‘hot-spots’’ where dominant primary emis-sions can sometimes overwhelm the influence of advection(Farrar et al., 2005a) (Fig. 6).
4.2. Derivation of estimated air concentrations
The pg/sampler data were used to derive estimated air con-centrations, as follows: (i) firstly, theoretical considerationswere used to determine whether a given compound was inthe linear (L), curvilinear (CL) or equilibrium (E) phase after7 days deployment time (see Table 1 for classification); (ii) forthe compounds in the L phase (i.e. compounds whereKOA > 10), an air sampling rate of 3 m3/day was assumedto apply (Harner et al., 2003), so the pg detected/samplerwas divided by 21 to give an estimate of the pg/m3 averagedair concentration for the 7 days deployment; iii. for the com-pounds in the E phase (i.e. compounds where KOA < 9), tem-perature corrected EVA:air partition coefficient data need tobe applied. Consequently, ambient temperatures for each sitewas applied to correct KOA data for the compound in question,from physico-chemical property measurements made by(Harner and Mackay, 1995; Harner and Bidleman,1996,1998; Harner and Shoeib, 2002). The semi-empirical
429N.J. Farrar et al. / Environmental Pollution 144 (2006) 423e433
Fig. 5. European HCB distribution, represented as pg/sampler. Biggest bar: 170 pg/sampler.
relationship between KOA and KEVA-AIR derived by Harneret al. (2003) was then used to back calculate air concentra-tions using eq. (1). Comparisons between measured and calcu-lated air concentrations for each compound class are given inTable 2, whilst other derived air concentration data can befound in Table 1.
4.2.1. PAHsThe derived air concentrations for S23PAHs varied between
1 and 63 ng/m3, with fluorene, phenanthrene, fluoranthene andpyrene dominating the congener profile. This is consistentwith the levels and composition of PAHs measured in air inactive sampling campaigns (Lee and Jones, 1999). The highest
Fig. 6. SPAH concentration across Europe (pg/sampler). Biggest bar: 28,100 pg/sampler.
European levels PASAE (mean) Backe et al., 2000 Lee et al.,
2004
Van Drooge et al.,
2002
Lee et al.,
2000
Harner et al.,
2000
Lee and
Jones, 1999
SPCB 270 (14e1700) 87 (41e161) e 78 (44e130) e 163 54e375
SPBDE 8 (1e50) e 12 (2.8e37) e e e e
SPCN 1220 (120e8170) e e e 152 59 e
HCB 410 (165e865) e e e 39 eLindane 411 (0e1776) e e e 500 (140e1070) e e
SPAH 10 (1e59) e e e e e 1.4e40
values were detected close to urban areas, at sites 4 (Russia)and 34 (Italy) for example, whiledas expecteddthe remotesites at 23 (Ireland) and 15 (Norway) gave the lowest values.High variability was seen within the compound groups, withfluoranthene, for example, ranging in air concentration bya factor of 88 between the sites.
4.2.2. PCBsThe derived air concentrations for PCBs also gave very
good agreement with measurements, varying by a factor of120 and falling in the 14e1700 pg S10PCB/m3 range. Table2 further illustrates the comparability between PCBs measuredusing POGs and active sampling data at an identical site innorthwest England. High levels were found in Germany(POG27; 1700 pg/m3) and Italy (POG32; 1600 pg/m3),whereas lower levels were found in coastal regions such asIceland (site 8; 45 pg/m3) and Norway (site 14; 14 pg/m3)(Fig. 8). Levels of variability between individual PCBs weremuch higher than PAHs. PCB-28, for example varies by a fac-tor of 138 between the highest and lowest sites, whilst PCB-138 varies by a factor of 800. Assuming that sources are thesame, this information provides a clue about their relativeLRAT potential.
4.2.3. PBDEsAs noted earlier, field blanks contained relatively high
levels of BDE-47, which hampered detection of that congenerin many samplers. In contrast, BDE-99 was detected in all thesamplers (Table 1). The derived air concentrations for this con-gener ranged from 25e410 pg/m3 (a factor of 16 variability),generally higher than other measurements (Strandberg et al.,2001; Farrar et al., 2004; Lee et al., 2004). Low levels weregenerally associated with remote coastal locations, such asNorway (site 15) and Eire (site 23), whilst higher levelswere often found in urban conurbations, such as Stockholm,Sweden (site 20) and Milan, Italy (site 34). The UK cities
were not particularly high (relative to other locations) in thisstudy, probably because cleaner air was coming from thewest throughout the sampling period.
4.2.4. PCNsAlthough levels of PCNs sequestered onto the POGs during
7 days of exposure (pg/sampler) were lower than that forPCBs, their lower KOA values resulted in lower effective airvolumes and consequently produced high air concentrationestimates for some lighter PCNs. Derived air concentrationsranged by a factor of 67 and were within the 35e2440 pg/m3
range. These values are high and may reflect uncertainties inthe KOA values. Harner and Bidleman (1997) found levelsbetween 24 and 175 pg/m3 (with an anomalously high valueof 469 pg/m3 which was attributed to localised burning), yetthis is still much lower than that seen for some individualPCN congeners here, for example (Table 1). The absolute con-centrations derived here should therefore be treated with somecaution at present.
4.2.5. Lindane g-HCHCalculated air concentrations ranged between less than the
detection limit and 1776 pg/m3. Highest values were in Portu-gal (site 17), France (site 25), Germany (site 28) and Spain(site 38). Comparisons between calculated and measured airconcentration data for lindane are very good. Previous work(Lee et al., 2000) has determined lindane concentrations ata semi-rural site in the UK in the 140e1070 pg/m3 range.Regions with current lindane usage would produce much higherlevels than the UK (where lindane is prohibited), particularlyif the pesticide was applied prior to or during deployment.
4.2.6. HCBLevels of this pesticide ranged from 170 to 870 pg/m3
(equivalent to a factor 5 variability), generally much higherthan that observed by other workers (Lee et al., 2000; Van
431N.J. Farrar et al. / Environmental Pollution 144 (2006) 423e433
Drooge et al., 2002). Concentrations in the UK have beenmeasured previously and are within the 38e76 pg/m3 range,yet UK samplers in this study detected levels equivalent to320e510 pg/m3. This difference may partly reflect underesti-mates in the air measurements made in some active samplingstudies, caused by HCB breakthrough.
4.3. Reasons for spatial variability
4.3.1. Air mass originAlthough some differences do emerge when comparing dis-
tribution plots (Figs. 5e8), there are also striking similarities.One is the generally low level of compounds sequestered bysamplers at coastal sites, particularly those impacted by airmasses originating from the Atlantic (e.g., Republic of Irelandand Portugal; sites 23 and 17 respectively) and Arctic Oceans(e.g., Iceland and Norway; sites 8 and 15 respectively). Backtrajectories indicate that prior to sampler contact, air massesimpacting coastal regions spent the vast majority of their jour-ney over the ocean (Fig. 4)dwhere air is generally low inPCBs and PBDEs (Lee et al., 2000, 2004). Once inland, airmasses entrain a myriad of compounds (from numerous sour-ces) that would potentially increase in concentration the longerit passed over land (i.e. as fetch increased). Consequently,samplers inland will tend sequester greater levels of POPs.
Having only sampled approximately 21 m3 of air, the abilityof the POG to distinguish between differences in air concentra-tions, even at geographically similar locations, is further illus-trated by comparing samples from Portugal (sites 16 and 17)and Spain (sites 35, 36, 37 and 38). Back trajectories indicatethat air masses influencing the Spanish samplers originatedfrom North Africa, whereas Portuguese levels are thought to
derive from Atlantic air masses. This is clearly exemplifiedin Fig. 7, as PBDEs in Southern Spain (Seville, site 37) areas much as 11 times greater than found in Portugal (site 13).
4.3.2. Population densityWorkers have quantified the existence of a rural-urban gra-
dient by carrying out local scale passive sampling campaigns,results of which strongly suggest that urban regions behave asa source for a range of POPs (Harner et al., 2004; Jaward et al.,2004). Related studies have ascribed regional differences in airconcentrations to previous production, as the synthesis ofPBDEs and PCBs was usually confined to a handful of coun-tries. To further credit these studies the possible relationshipbetween urbanisation and air concentration was explored byoverlaying the amount of compound sequestered per samplerwith population density data.
With greater degrees of population density evidentlylocated in Central and Western Europe, it would be expectedthat compounds associated with human activity would pre-dominate. This is the case, as greatest levels of PBDEs wereseen at sites in Germany (site 27; 1000 pg/sampler) and Italy(site 34: 800 pg/sampler), with low levels in remote regions(e.g., 65 pg/sampler at site 15; Tromso, Norway). In contrastto remote regions, urban areas will be impacted more fromlocal sources than advection. PAHs, already seen to be locatedin urban hot-spots, also appear to be associated with increasedpopulation density.
4.4. Correlation of PCNs against PAHs and PCBs
A correlation matrix can be utilised as a means of establish-ing a possible relationship between two datasets. In this
Fig. 7. Spatial variation of SPBDEs across Europe, represented as pg/sampler. Biggest bar: 760 pg/sampler.
Fig. 8. Spatial variation of SPCBs across Europe, represented as pg/sampler. Biggest bar: 4770 pg/sampler.
instance, levels of PCNs (pg/sampler and pg/m3) were corre-lated against both PAHs and PCBs. Workers have previouslysuggested that some PCN congeners may be linked to combus-tion sources (e.g. CN-29, -51, -52/60, -54 and -66/67) (Meijer,2001). However, these congeners were not detected in thesamples. Correlating the PCNs detected in this study (Table. 1)with PAHs (compound with a primary combustion source)resulted in only one PAH (benzo[ghi]perylene) being signifi-cantly correlated at the p < 0.01 level. In contrast, a correlationbetween PCNs and PCBs produced a significant correlation,with most congeners correlated at the p < 0.05 or p < 0.01levels.
References
ApSimon, H.M., Warren, R.F., Kayin, S., 2000. Addressing uncertainty in
environmental modelling: a case study of integrated assessment of strategies
to combat long-range transboundary air pollution. Atmospheric Environ-
ment 3, 5417e5426.
Backe, C., Larsson, P., Okla, L., 2000. Polychlorinated biphenyls in the air of
southern Swedendspatial and temporal variation. Atmospheric Environ-
ment 34, 1481e1486.
Bailey, R.E., 2001. Global hexachlorobenzene emissions. Chemosphere 43,
167e182.
Barber, J.L., Thomas, G.O., Kerstiens, G., Jones, K.C., 2003. Study of plant-
air transfer of PCBs from an evergreen shrub: implications for mechanisms
and modelling. Environmental. Science and Technology 37, 3838e3844.