Environ Sci Pollut Res (2015) 22:11301–11313 DOI 10.1007/s11356-015-4363-4 RESEARCH ARTICLE Air and seawater pollution and air–sea gas exchange of persistent toxic substances in the Aegean Sea: spatial trends of PAHs, PCBs, OCPs and PBDEs Gerhard Lammel1,2 & Ondřej Audy1 & Athanasios Besis3 & Christos Efstathiou1 & Kostas Eleftheriadis4 & Jiři Kohoutek1 & Petr Kukučka1 & Marie D. Mulder1 & Petra Přibylová1 & Roman Prokeš1 & Tatsiana P. Rusina1 & Constantini Samara3 & Aysun Sofuoglu5 Sait C. Sofuoglu5& & Yücel Taşdemir6 & Vassiliki Vassilatou4 & Dimitra Voutsa3 & Branislav Vrana1 Received: 19 November 2014 /Accepted: 11 March 2015 /Published online: 25 March 2015 # Springer-Verlag Berlin Heidelberg 2015 Abstract Near-ground air (26 substances) and surface seawa- DDT and seawater concentrations of p,p′-DDE and p,p′- ter (55 substances) concentrations of persistent toxic sub- DDD were elevated in Thermaikos Gulf, northwestern stances (PTS) were determined in July 2012 in a coordinated Aegean Sea. The polychlorinated biphenyl (PCB) congener and coherent way around the Aegean Sea based on passive air pattern in air is identical throughout the region, while (10 sites in 5 areas) and water (4 sites in 2 areas) sampling. polybrominated diphenylether (PBDE)patterns are obviously The direction of air–sea exchange was determined for 18 PTS. dissimilar between Greece and Turkey. Various pollutants, Identical samplers were deployed at all sites and were polycyclic aromatic hydrocarbons (PAHs), PCBs, DDE, and analysed at one laboratory. hexachlorobenzene (HCB), hexa- penta- and hexachlorobenzene are found close to phase equi- chlorocyclohexanes (HCHs) as well as dichlorodiphenyltri- librium or net-volatilisational (upward flux), similarly at a chloroethane (DDT) and its degradation products are evenly remote site (on Crete) and in the more polluted Thermaikos distributed in the air of the whole region. Air concentrations of Gulf. The results suggest that effective passive air sampling p,p′-dichlorodiphenyldichloroethylene (p,p′-DDE) and o,p′- volumes may not be representative across sites when PAHs significantly partitioning to the particulate phase are included. Responsible editor: Philippe Garrigues Electronic supplementary material The online version of this article Keywords Air–sea gas exchange . Deposition . Passive (doi:10.1007/s11356-015-4363-4) contains supplementary material, sampling . Volatilisation . Fugacity ratios . Aegean Sea which is available to authorized users. * Gerhard Lammel lammel@recetox.muni.cz Introduction 1 Research Centre for Toxic Compounds in the Environment, Masaryk Persistent toxic substances (PTS) pose a hazard for ecosys- University, Brno, Czech Republic tems and human health as they undergo long-range atmo- 2 Multiphase Chemistry Department, Max Planck Institute for spheric transport (LRT), are ubiquitous in the global environ- Chemistry, Mainz, Germany ment and, through bioaccumulation along food chains, may 3 Department of Chemistry, Environmental Pollution Control reach harmful levels even in remote areas (UNEP 2003;WHO Laboratory, Aristotle University, Thessaloniki, Greece 2003). PTS include organochlorine pesticides (OCPs), such as 4 Institute of Nuclear Technology and Radiation Protection, NCSR hexachlorobenzene (HCB), hexachlorocyclohexane (HCH), Demokritos Institute, Athens, Greece and dichlorodiphenyltrichloroethane (DDT), and banned in- 5 Department of Chemical Engineering, Izmir Institute of Technology, dustrial chemicals such as polychlorinated biphenyls (PCBs), Urla, Izmir, Turkey banned in Europe since the 1970s, only recently restricted 6 Environmental Engineering Department, Uludağ University, industrial chemicals, such as polybrominated diphenylethers Nilüfer, Bursa, Turkey (PBDEs), and combustion by-products, such as polycyclic 11302 Environ Sci Pollut Res (2015) 22:11301–11313 aromatic hydrocarbons (PAHs, all combustion types) and also sites in central and northern Greece and in western Turkey, HCB (waste burning), which primary emissions are ongoing. as well as at one urban site in Thessaloniki, northern Greece. LRT of chlorinated PTS (Semeena et al. 2006; UNECE 2010) Surface seawater was sampled at one of the Cretan sites and at and also of PAHs (Lammel et al. 2009; Galarneau et al. 2014) three sites in the Thermaikos Gulf, northern Aegean Sea is enhanced by re-volatilisation from surfaces (multihopping; (Table 1). significance of secondary sources in air). The same can be expected for most other PTS, as they resist to degradation in Sampling the soil and surface water and are semivolatile (vapour pres- sures between 10−6 and 10−2 Pa at 298 K). Environmental fate Air monitoring is needed for global PTS assessment (Klánová et al. 2011) and part of the National Implementation Plans of PTS were collected by PAS using polyurethane foam (PUF) parties to the Stockholm Convention (UNEP 2014), including disks (Molitan, Gumotex, Czech Republic; density of Greece and Turkey. High PTS levels in air and seawater are a 0.030 g cm−3, 150 mm diameter, 15 mm thickness): Before concern in the Mediterranean environment (Lipiatou and use, PUF disks were cleaned through 8-h Soxhlet extraction Saliot 1991; Kouimtzis et al. 2002; UNEP 2002; Terzi and with acetone and dichloromethane (DCM) and placed in a Samara 2004; Biterna and Voutsa 2005; Mandalakis et al. glass cartridges. PUF disks were deployed in protective cham- 2005; Tsapakis et al. 2006; Chrysikou et al. 2008; Chrysikou bers consisting of two stainless steel bowls (upper 30 cm di- and Samara 2009). The Mediterranean region is characterised ameter and lower 24 cm diameter; Klánová et al. 2008). by strong urban and industrial sources and adjacent source PAS were deployed during 28–30 days at each site (see regions (western, central and eastern Europe). Exposure to Table 1), except at Selles Beach (11 days, 2–13 July 2012) long-range transported pollution from central and eastern and at Nilüfer-Bursa (42 days, 2 July–13 August 2012). All Europe is highest in summer (Lelieveld et al. 2002). PAS data are corrected for field blanks. Between one and four Passive air sampling (PAS) is most useful for PTS moni- field blanks were produced at each sites by leaving the PUF toring and mapping, although the uncertainty of concentration disks exposed within the PAS for a few seconds and keeping measurements is within a factor of 2–3 (Harner et al. 2006; all other operations identical to sample processing. Values Klánová et al. 2008). Trace contaminants in water can be below the mean of the field blank values, b, plus three relative quantified using hydrophobic passive water samplers (PWS) standard deviations (σ) of the field blank values were consid- with an uncertainty of about a factor of 2 (Lohmann et al. ered 10,000. For BDE 209, the MS resolution was set to >5000. Injection was splitless 2μL at 280 °C, withHe as Air–sea diffusive gas exchange calculations carrier gas (1 mL min−1). The GC temperature programme was 80 °C (1 min hold), then 20 °C min−1 to 250 °C, followed State of phase equilibrium is addressed by fugacity calcula- by 1.5 °C min−1 to 260 °C (2 min hold) and 25 °C min−1 to tion, based on the Whitman two-film model (Liss and Slater 320 °C (4.5 min hold). For PCBs and OCP GC-MS/MS anal- 1974; Bidleman and McConnell 1995). The fugacity ratio ysis was performed on a 7890A GC (Agilent, USA) equipped (FR) is calculated as: with an HT8 column (SGE, USA) and coupled to a 7000BMS   (Agilent, USA), operated in EI+MRM. Injection was splitless FR ¼ f a= f w ¼ caRT a= cwHTw;salt ð2Þ 3μL at 280 °C, with He as carrier gas (1.5mLmin−1). The GC with gas-phase concentration c (ng m−3), dissolved aqueous temperature programme was 80 °C (1 min hold), then a −1 −1 concentration cw (ng m −3), universal gas constant R 40 °C min to 200 °C, and finally 5 °C min to 305 °C. (Pa m3 mol−1 K−1), water temperature and salinity corrected Henry’s law constant H (Pa m3 mol−1) and air tempera- Quality assurance/quality control Recoveries of PAHs in Tw,salt ture Ta (K), which was measured continuously on site. ThePASwere 79–129 %, except for dibenz(a,h)anthracene, which water temperature was assumed to be 0.5 K lower than the air was 150 % in PASs and 61–101 % in SRs, except for ace- temperature, throughout, based on discontinuous measure- naphthylene, anthracene and benzo(a)pyrene, which were 29, ments at site 1b, 10–30 cm below the surface, which covered 53 and 49 %, respectively. Recoveries of PCBs and OCPs morning, noon, evening and night time measurements. Values from PASs were 80–88 % and 71–114 %, respectively, while 0.33.0 indicates net deposition and OCP results, but were applied for all PBDE results. Recovery FR<0.3 net volatilisation. The diffusive air–seawater gas ex- of surrogates (deuterated substances, see above) varied from change flux (F , ng m−2 day−1) is calculated according to the 88 to 100 % for PCB and from 72 to 102 % for PAHs. Field aw Whitman two-film model (Bidleman and McConnell 1995; blanks for various types of sample were collected along with Schwarzenbach et al. 2003): the samples. For each site separately, the mean of one to four   PAS field blank values was subtracted from the air sample Faw ¼ kol cwcaRT a=HTw;salt ð3Þ values. Values below the mean plus three standard deviations of with air–water gas exchange mass transfer coefficient kol −1 the field blank values were considered below limit of quanti- (m h ), accounting for resistances to mass transfer in both −1 −1 fication (LOQ). Field blank values of most analytes in air water (kw, m h ) and air (ka, m h ) (see Bidleman and samples were lower than the instrument limits of quantifica- McConnell 1995; Zhong et al. 2012; and references therein). tion (ILOQ) at most sites. Highest LOQs for analytes in PASs, Substance property data are taken from the literature (refer- up to 0.028 and 0.13 ngm−3, resulted for acenaphthene (ACE) ences listed in the SM, S1). and fluorene (FLN), respectively, and up to 24, 8.6 and Sampler uptake kinetics and the half-time of equilibration 1.2 pg m−3 resulted for PeCB, β-HCH and γ-HCH, respec- of organics dissolved in water in SRs is substance dependent. tively. Higher LOQs for analytes in the high-volume AASs, For most substances the derived concentrations reflect the 11306 Environ Sci Pollut Res (2015) 22:11301–11313 mean over the entire exposure period, but for HCH and some Σ5PAHs two- to three-ring PAHs only the mean of the last 12–24 days Gülbahçe, Izmir of the exposure period. Substances which, accordingly, were Nilüfer, Bursa determined in air and water simultaneously only for a short Polikos, EuboeaAthens, Demokritos period (air sampling terminated 14 days before water sam- Thermaikos Gulf, Loudias River estuaries Thermaikos Gulf, Michaniona pling), i.e. HCH isomers, NAP, ACE, ACY and FLN were Thessaloniki, Agia Sophia excluded from diffusive air–sea exchange calculations. NeochoroudaSelles Beach, Crete Finokalia, Crete 0,0 1,0 2,0 3,0 4,0 Results and discussion ng m-3 Air mass origin Σ7PCBs The predominant origin of air masses during the sampling Gülbahçe, Izmir Nilüfer, Bursa campaign was central, eastern and southeastern Europe. This Polikos, Euboea is shown as the distribution of residence time of air masses in Athens, Demokritos Thermaikos Gulf, Loudias River estuaries Fig. S1. This analysis is based on re-analysis data and back Thermaikos Gulf, Michaniona trajectory modelling (FLEXPART model; Stohl et al. 1998). Thessaloniki, Agia Sophia Neochorouda During 2–11 July 2012 the Aegean was mostly influenced Selles Beach, Crete by northerly, in its northern part easterly advection as part of a Finokalia, Crete cyclonic system, which moved from western Russia to 0 50 100 pg m-3 Romania. Under the influence of a strong westerly flow to- wards Europe, the flow in the northern part of the Aegean switched to westerly during the night 11–12 July, such that Σ9OCPs air which had been residing over the SW Balkans was Gülbahçe, Izmir Nilüfer, Bursa advected, as well as air from beyond, i.e. central Italy, the Polikos, Euboea NW Mediterranean Sea and the Iberian Peninsula. Air from Athens, Demokritos Thermaikos Gulf, Loudias River estuaries central Europe flew to the Aegean until the night of 17–18, Thermaikos Gulf, Michaniona and from the northeast thereon. Then, for a few days, air Thessaloniki, Agia Sophia Neochorouda arrived in the Aegean from central Europe, directed Selles Beach, Crete (counter-clockwise) around a cyclonic system above Italy. Finokalia, Crete From the evening of 27 July until the beginning of August, 0 50 100 -3 eastern flow arrived in the Aegean. The mean air temperatures pg m ranged 26–29 °C in the study area with higher temperature maxima at the Thessaloniki and Athens sites (Table 1). No Σ5PBDEs precipitation occurred during the measurement period with Gülbahçe, Izmir only one exception, namely on 30 July at the Thermaikos Nilüfer, BursaPolikos, Euboea Gulf sites. Athens, Demokritos Thermaikos Gulf, Loudias River estuaries Thermaikos Gulf, Michaniona Concentration levels Thessaloniki, Agia Sophia Neochorouda Selles Beach, Crete Mean concentrations of PAHs, PCBs, OCPs and PBDEs Finokalia, Crete found in near-ground air and surface seawater are shown in 0 2 4 6 8 10 12 14 16 pg m-3 Figs. 1, 2 and 3. Concentrations of the individual substances targeted are provided in Tables S4–5. Fig. 2 Sum concentrations of PAHs, PCBs, OCPs and PBDEs found in the air at various sites across the Aegean Sea Levels, trends and patterns of PTS in near-ground air Polycyclic aromatic hydrocarbons The significance of local highest levels occurring at urban and residential sites in the primary sources and, hence, spatial variation is higher for Thessaloniki area (Figs. 1 and 2, Tables 2, S4–S5). PAH pat- PAHs than for halogenated PTS (Lammel et al. 2010; terns in air seem to be quite dissimilar across sites (mean Lammel 2015). In this study, we found a trend of increasing correlation coefficient r=0.60; Table S6a). However, because PAH levels from remote to rural and residential sites, with the of major PAHs commonly found not being included in this Environ Sci Pollut Res (2015) 22:11301–11313 11307 Σ27PAHs and related compounds analysis (but only five substances), a conclusion on source contributions would not be justified (Dvorská et al. 2012). Thermaikos Gulf, mussel culvaons Indeed, the similarity across sites is high (mean r=0.93) if PAH patterns are compared across sites on a pg basis (i.e. Thermaikos Gulf, Loudias River estuaries PAS results not divided by effective sampling volume; Thermaikos Gulf, Michaniona Table S7 for 10 PAHs). Then, just some dissimilarity from the prevailing pattern is indicated for one site, Bursa (r= Selles Beach, Crete 0.74–0.81). This can be explained by the inclusion of sub- stances significantly partitioning to the particulate phase in 0 2000 4000 6000 the targeted substance class, namely FLT and PYR (as shown pg L-1 in the region; Terzi and Samara 2004, besides others). Gas- particle partitioning of semivolatiles, and hencethe effective Σ7PCBs sampling volume, depends not only on temperature (similar across sites) but also on the particulate phase chemical prop- erties, which was certainly dissimilar across site types and Thermaikos Gulf, mussel culvaons across sites of same type. This is related to the processes Thermaikos Gulf, Loudias River estuaries determining gas-particle partitioning of semivolatiles, which are dependent on particulate phase chemical properties, such Thermaikos Gulf, Michaniona as organic and black carbon abundances (Lohmann and Selles Beach, Crete Lammel 2004). The PAH levels at the remote sites 1a–b are similar to those observed at a high mountain site in 2006 0 20 40 (Moussala, SW Bulgaria) and in the open eastern pg L-1 Mediterranean Sea in 2010, but lower than those observed in the Aegean Sea, eastern Mediterranean Sea, and Black Seas in 2006 (ship measurements; Table 2). Local primary sources Σ10OCPs should also explain the levels found in Aliartos, central Greece, in 2006 and a suburban site in the Izmir area in Thermaikos Gulf, mussel culvaons 2003–2004, which were more than one order of magnitude higher (Table 2). The concentrations reported from rural sites Thermaikos Gulf, Loudias River estuaries in the Republic of Macedonia (Stafilov et al. 2011; Table 2) Thermaikos Gulf, Michaniona are based on the same sampling and analysis methods as our study, but V28d=100 m 3 was assumed throughout, which is up Selles Beach, Crete to a factor of 6 below values of V28d used in here. This may 0 200 400 600 explain the high levels reported from there. pg L-1 The levels of PAHs at the urban/residential sites in the area of Thessaloniki (2a–c) are low among the range of levels reported for urban and rural sites in Greece for gaseous Σ8PBDEs PAHs collected by active sampling (Manoli et al. 2011, and references herein). For PAHs and HCHs, the differences span Thermaikos Gulf, mussel culvaons more than one order of magnitude in some cases (Table 2). Partly, this discrepancy might be explained by overestimated Thermaikos Gulf, Loudias River estuaries sampling efficiencies (discussed in section S2). Thermaikos Gulf, Michaniona Polychlorinated biphenyls PCBs in the Thessaloniki, Selles Beach, Crete Athens and Izmir areas (sites 2a, 3a and 5), as well as in the Bursa area, NW Turkey, in 2008–2009 (Birgül and Taşdemir 0 5 10 2011), are found (only) a factor of 2 higher than at the remote pg L-1 sites 1a and 1b. This points to the significance of PCB sources in urban areas for regional distribution discussed previously Fig. 3 Sum concentrations of PAHs and related compounds, PCBs, (Diamond et al. 2010). Urban-to-rural PCB gradients were OCPs and PBDEs found in surface sea water at coastal sites across the also observed in the Mediterranean (Mandalakis et al. 2002; Aegean Sea Gasić et al. 2010). PCB patterns in air are almost identical across all sites in the region (Table S6a). The same is found 11308 Environ Sci Pollut Res (2015) 22:11301–11313 Table 2 Overview of contaminants’mean concentrations observed in near-ground air of the region in recent years (PAHs, ng m−3; all other, pg m−3). Limited number of individual substances included for the sake of comparability Site, year, reference ΣPAH a5 ΣPCB b 7 HCB ΣHCH c ΣDDTsd ΣPBDE e3 Remote Crete, remote (summer 2012, this work, sites 1a–b) 0.070–0.63 5.8–25.9 8.1–19 3.4–15.8 3.2–6.3/0.40–0.78 0.4–0.78 Crete, remote (summer 2006; Iacovidou et al. 2009) 8.4f 1.2 Thessaloniki area, coastal remote (summer 2012, this work, site 2d) 0.76 22.6 11 11.5 13.8 SW Bulgaria, remote, high mountain (Moussala, summer 2006; Halse et al. 2011) 0.63 10.3 55.9 19.7 Aegean Sea (cruise, summer 2006; Castro-Jiménez et al. 2012; Berrojalbiz et al. 2014) 5.5f 69–234 11–106 39–85 Black Sea (cruise, summer 2006; Castro-Jiménez et al. 2012) 7.4g C and E Mediterranean (cruises, June 2006, May 2007; Castro-Jiménez et al. 2012; 4.7 29–234 5–178 <3–155 Berrojalbiz et al. 2014) E Mediterranean (cruise summer 2010; Mulder et al. 2013, 2014) 0.61 3.3 6.0 1.4 4.3 Rural, residential Thessaloniki area, residential (summer 2012, this work, sites 2a, 2c) 2.33–3.64 26.7–47.5 9.9–10 9.3–10.4 8.4–51/0.56–1.12 Athens, suburban (summer 2012, this work, site 3a) 0.25 50.2 10.2 9.7 4.7 1.5 Central Greece, rural (summer 2012, this work, site 3b) 0.58 9.0 6.7 4.8 7.0 0.9 Izmir, suburban, W Turkey, whole year 2003–2004 (Demircioglu et al. 2011) 21.9 W Greece, two rural sites, whole year 2000–2001 (Terzi and Samara 2004) 2.9–13.8j C Greece, rural (Aliartos, summer 2006; Halse et al. 2011) 23.7 13.6 54.0 228 NW Turkey, semi-residential (summer 2012, this work, site 4) 0.23 25.6 5.5 5.7 7.9 0.7 NW Turkey, semi-rural, full year 2008–09 (Birgül and Taşdemir 2011) 74 NW Turkey, rural, coastal, full year 2008–2009 (Yolsal et al. 2014) 80 W Turkey, rural (summer 2012, this work, site 5) 0.27 55.3 12.3 13.8 12.8 8.5 Macedonia, two rural sites, summer 2007 (Stafilov et al. 2011) 24–174 37–46 101–182 36–126 Urban, industrial Thessaloniki area, urban (summer 2012, this work, site 2b) 2.38 85.7 9.8 14.9 42.5/6.96 W Greece, urban site, whole year 2000–2001 (Terzi and Samara 2004) 21.7h Athens, urban, December 2006 (Mandalakis et al. 2009) 3.6 Two sites in the suburban area of Athens, June and November 2003 17.5–20.1h (Vasilakos et al. 2007) Iraklion, semi-urban, 2006–2007 (Mandalakis et al. 2009) 10.7 Republic of Macedonia, four urban sites, summer 2007 (Stafilov et al. 2011) 74–278 35–51 165–3033 106–246 Izmir, W Turkey, winter/summer 2005 (Pozo et al. 2009) 644/287 29/48 51/60 Izmir, W Turkey, winter 2004 (Demircioglu et al. 2011) 100 Izmir, W Turkey, spring 2003 (Sofuoglu et al. 2004) 228 49 Aliaga, W Turkey, all seasons 2009–2010 (Kaya et al. 2012) 78 2560 Zonguldak, N Turkey, winter/summer 2007–2008 (Akyüz and Çabuk 2010) 260/21 Konya, S Turkey, full year 2006–2007 (Ozcan and Aydin 2009) 93 78 520 130 Bursa, NW Turkey, all seasons 2004–2005 (Taşdemir and Esen 2007) 20 Bursa, NW Turkey, full year 2008–2009 (Birgül et al. 2011; Yolsal et al. 2014) 117h 61 a Sum of ACE, FLN, PHE, FLT and PYR b PCB28, PCB52, PCB101, PCB118, PCB153, PCB138 and PCB180 c Sum of α- and γ-HCH d Sum of DDT and DDE isomers e Sum of BDE47, BDE99 and BDE100 fWithout PCB28 and PCB138 gWithout ACE hWithout ACE and FLN with slightly lower correlation coefficients if PCB the congeners (Mandalakis et al. 2003) would suggest patterns are compared across sites on a pg basis that the pattern changes during transport from (urban) (Table S7). Selective photochemical degradability of sources to the remote site, not confirmed here. Environ Sci Pollut Res (2015) 22:11301–11313 11309 Organochlorine pesticides The levels of HCB, HCHs and Polybrominated diphenylethers BDE levels in air are found DDX in air at the rural and residential sites 3a–b, 4 and 5 at sites 2b (urban), 2d (remote coastal) and 5 (rural) and are are not elevated against the remote sites, but similar. The found about one order of magnitude higher than at the other highest concentrations of HCB and α-HCH are actually found sites (Figs. 1d and 2, Tables S4a, S5). While the high levels at the remote site 1a (Tables S4a, S5). The ratio α-HCH/γ- correspond to what was reported in 2006–2007 from Greek HCH ranges widely, from 1.3 to 7.8. The lowest values of this urban and semi-urban sites (Mandalakis et al. 2009; Table 2), ratio are observed at the urban sites 2a, b, close to what was the lower levels (sites 1a, 1b, 2a, 2c, 3b and 4; Fig. 2, previously found at site 2b in summer for the particle-bound Tables S4a, S5) are below earlier observations, including HCH isomers (Chrysikou and Samara 2009). The maximum one from Finokalia, i.e. site 1a (Iacovidou et al. 2009; α-HCH/γ-HCH ratio is observed at the remote site 1a (re- Table 2). The measured concentrations might be mote, 7.8) and may reflect the effect of the higher Henry underestimated due to overestimated sampling efficiencies coefficient (units of Pa m3 mol−1) of α-HCH in an environ- (discussed in section S2) for some sites (2a, 2b, 3a and 4). ment where the atmospheric levels are dominated by the Surprisingly, while at most Greek sites, the most abundant source re-volatilisation from surface seawater. OCP patterns congener was BDE47 followed by BDE99; it was BDE28 at in air are found highly correlated across all sites except the the Greek site 3b and at the two Turkish sites. PBDE patterns remote site 1a (also when compared on a pg basis, see in air are obviously similar among most Greek sites (even Tables S6a and 7), which show different patterns, e.g. DDT/ identical, r=1.0 between sites 1 and 2, and sites 2 and 3), DDX is the highest. This may also indicate substance selective but dissimilar among the Turkish sites and across the countries sinks (dry deposition, photochemistry) or sources (re- (Table S6a). When PBDE PAS results are expressed as ng volatilisation from the sea surface) active during transport rather than as ng m−3 (Table S7), correlation coefficients are from continental sources to Crete. The prevailing spatial ho- not higher, but lower. This is noteworthy, as the targeted mogeneity of the concentration levels confirms the perception PBDEs are significantly partitioning to the particulate phase of long-lived pollutants undergoing even distribution within a (Chen et al. 2006; Cetin and Odabasi 2008; Su et al. 2009), region, or even being dominated by LRT from outside the similar to targeted PAHs, for which correlation coefficients are region. In general, they are lower than reported previously higher when based on ng (above). Observations suggest that from the same type of sites in the region (though only few PBDE phase partitioning is largely determined by adsorption measurements available; Table 2). At the remote sites, HCB (Cetin and Odabasi 2008; Su et al. 2009). This suggests that and DDX levels are found similar to ship measurements in different aerosol chemical composition (across sites) could 2010 in the eastern Mediterranean (Mulder et al. 2013), while hardly affect partitioning, in agreement with the finding here DDX was much lower than at a remote high mountain site in (Tables S6a and S7). However, predictability of PBDE SW Bulgaria (Halse et al. 2011; Table 2). Furthermore, DDX partitioning is low, as most congeners are not expected to be had been reported very high at a rural site in central Greece, in phase equilibrium as a consequence of high Koa values Aliartos, in 2006 (Halse et al. 2011; Table 2). (Cetin and Odabasi 2008). Therefore, phase partitioning of A fairly consistent regional distribution with rather individual congeners could vary across sites and in different small urban-to-rural/residential concentration gradients air masses at the same site even for similar temperatures and was found for HCB, HCHs and DDX, and small rural/ similar particulate phase chemical properties, determined by residential-to-remote gradients for HCB and HCHs. aging. Similar conclusions were drawn for HCB and HCHs based on PAS across the entire European continent in Levels and trends in surface seawater 2006–2007 (Halse et al. 2011). I n c o n t r a s t , D D X s , n a m e l y p , p ′ - Mean levels of the sum concentrations of PAHs and related dichlorodiphenyldichloroethylene (p,p′-DDE) and also o,p′- compounds, PCBs, OCPs and PBDEs found in surface sea- DDT, were elevated in air at the urban and coastal sites in water at four coastal sites across the Aegean Sea are presented the Thermaikos Gulf (2b, 2c and 2d, Fig. 2, Table S4a), as in Fig. 3 (for individual substances targeted, see Table S4). well as in seawater (namely p,p′-DDE and p,p′-DDD, sites 2c and 2d, Fig. 2, Table S4b). The fraction of parent DDTamong Pollutant levels and spatial trends The concentrations of the DDX compounds is low in both air and seawater (ranges PAHs, PCBs, HCHs and DDX compounds in surface seawa- 0.03–0.43 and 0.01–0.37, respectively). This indicates the ab- ter of the Thermaikos Gulf are a factor of 2–10 higher than at sence of any fresh DDT input to themarine environment of the the Cretan site, 1b (Figs. 1 and 3, Table S4b). This concentra- Aegean. DDT had been banned in the 1970s and 1980s tion gradient is exceeding one order of magnitude for some (Pacyna et al. 2003). However, the surface seawater of the PAHs, notably PHE, RET, BBN, BGF, CHR and BEP. This eastern Mediterranean is expected to be a source for DDT may reflect direct discharges from nearby urban activities and since the 1980s (Stemmler and Lammel 2009). from ships, or photochemical degradation during atmospheric 11310 Environ Sci Pollut Res (2015) 22:11301–11313 transport to the remote site. A similar north–south gradient of consistent across all sites (Table S6b). PCB patterns in seawa- PCB was observed in 2006–2007, with a concentration span ter are consistent, too, but less than in air (Table S6a, b). of two to four between the northern and southern Aegean Sea However, there is also a difference in the PCB pattern along (dissolved fractions; Berrojalbiz et al. 2011). the off-shore gradient. The OCP pattern at the remote site 1b is For PeCB, HCB and PBDEs, no significant difference in very different from the patterns found in the Thermaikos Gulf. seawater concentrations is found across sites. The ratio α-HCH/γ-HCH ranges 0.4–2.2 in seawater, lower The spatial variation of seawater pollution across the three than in air. This is in line with the difference in Henry coeffi- Thermaikos Gulf sites is considerable for PAHs (highest at the cients higher (in units of Pa m3 mol−1) for α- than γ-HCH. residential coastal site 2c, Michaniona) and DDX (highest at Along the off-shore gradient α-/γ-HCH almost triples from the remote site 2d, Loudias River estuaries; Figs. 1 and 3, 0.4 to 1.1, indicating minimal values in freshwater of the area. Table S4b). The highest OCP levels at the Loudias River This might be related to air–sea exchange of the isomers, not plume imply influence from agricultural activities, whereas effective in or close to the estuary. the highest levels of PCBs and PBDEs at the off-shore site 2e (located 1.4 km from the coastal site 2d) may indicate a shift of contaminant partitioning from particle-bound to dis- Air–sea gas exchange solved, corresponding to a negative gradient of suspended particulate matter concentration, or influence from shipping The direction of air–sea exchange was derived for three sites in activities. two areas, 1b (Crete), 2c and 2d (Thermaikos Gulf). The three- ring PAHs ACE, FLN and PHE are found volatilisational (i.e. Comparison with previous observations Our data represent upward net flux) in the Thermaikos Gulf (no measurement at a snapshot in time. The PAH levels in the Thermaikos Gulf are the Cretan coast) (Fig. 4). Such a result is not unexpected for significantly lower than those found in the northern Aegean coastal waters in the vicinity of strong primary emissions and Sea in 1997 (10–30 ng L−1, however, unclear whether total or had been observed before in coastal waters of the northeastern dissolved concentrations; UNEP 2002). PCB concentrations USA (Lohmann et al. 2011) and even in the open southeastern in two samples of seawater collected in the southern Aegean Mediterranean Sea in spring 2007 (Castro-Jiménez et al. 2012). Sea (close to Crete) was lower in 2006–2007, ΣPCB7=2.6 Some three- to four-ring parent PAHs, among them FLN, had and 3.7 pg L−1, in contrast to 7.5 pg L−1 at the Cretan site in been reported to be close to phase equilibrium in the 2012. PCB in two seawater samples collected in the northeast- Mediterranean and Black Seas (Castro-Jiménez et al. 2012). ern and central Aegean Sea 2006–2007 was somewhat lower, Penta- and hexachlorinated PCBs are found with 7.3 and 13.8 pg L−1 (Berrojalbiz et al. 2011), than that found in volatilisational trends at site 2c, Michaniona in the 2012 in the Thermaikos Gulf (19.6–33.4 pg L−1; Table S4b). Thermaikos Gulf, and all seven indicator PCBs, except Dissolved ΣPCB7 was also lower, 13.1 pg L −1, in 1987 in PCB118, at site 1b, Selles Beach on Crete. They seem to be surface seawater of the southern Aegean, Cretan and Ionian close to phase equilibrium at site 2d in the Thermaikos Gulf. Seas (Schulz-Bull et al. 1997). HCB is found consistently PCB were concluded to be net volatilisational in the eastern higher by a factor of about 5 than in the northern Aegean Mediterranean based on measurements in air in 2001–2002 Sea in 2006–2007 (dissolved phase; Berrojalbiz et al. 2011). and in seawater in the 1990s (18 congeners; Mandalakis et al. β- and γ-HCH concentration at site 1b are also higher than in 2005) but close to phase equilibrium in the Aegean Sea in 2006–2007, namely by a factor of 3 and 5, respectively. 2006–2007 (41 congeners; Berrojalbiz et al. 2014). HCB Concentration levels of OCPs, PCBs and PAHs in seawater and PeCB are found net volatilisational at all three sites. of the Thermaikos Gulf have exhibited a considerable de- HCB had been observed but close to phase equilibrium in crease during the last two decades (Kamarianos et al. 2002). the Aegean Sea in 2006–2007 (Berrojalbiz et al. 2014). The Nevertheless, several PTSs have been recently detected in direction of air–water gas exchange of HCH isomers is found sediments from Thermaikos Gulf including p,p′-DDE, α- depositional in the Thermaikos Gulf (Fig. 4). This is, to our and β-HCH, PCBs (mainly the congeners 138, 101, 28 and knowledge, the first observation of pentachlorobenzene fu- 180; Terzopoulou and Voutsa 2011), PBDEs (Dosis et al. gacities in European marine environments. 2011), and PAHs (IKYDA 2010). Among the PBDEs only BDE28 seems to approach phase equilibrium (fa/fw=1–20 at the three sites; Fig. 4) while the Contaminant patterns The ratio DDT/DDX, indicative for other congeners tested, BDE47, BDE66, BDE100 and aging, equals 0.37 in seawater of the Cretan coast, site 1b, and BDE99, were found to be depositional with fa/fw in the range much less, 0.01–0.05, in the Thermaikos Gulf. This may point 102–104 (BDE66 and BDE100 at all sites, BDE47 at site 1b to some influence of long-range transported DDT in the south- on Crete) or even higher (BDE99 at all sites, BDE47 at the ern Aegean Sea as opposed to the Thermaikos Gulf environ- Thermaikos Gulf sites), i.e. very far from phase equilibrium ment. PAH and PBDE patterns in seawater are highly (not included in Fig. 4). Environ Sci Pollut Res (2015) 22:11301–11313 11311 Fig. 4 Fugacity ratios, fa/fw, at sites on Crete (1b Selles Beach) and in the Thermaikos Gulf (2c Michaniona, 2d Loudias River estuaries). Only substances with Cw>LOQ considered. Ca