Feasibility of using low density polyethylene sheets to detect Feasibility of using low density polyethylene sheets to detect atmospheric organochlorine pesticides in Alexandria, Egypt atmospheric organochlorine pesticides in Alexandria, Egypt

14 15 Egypt is a major agricultural country in Africa with a known past of organochlorine pesticides 16 (OCPs) application, yet data on atmospheric levels of OCPs in Egypt is sparse. Low density 17 polyethylene (LDPE) passive samplers were therefore deployed for 3 weeks each at 11 locations 18 in July, 2010 and January, 2011 in Alexandria to screen for gas-phase OCPs. Performance 19 reference compounds were used to investigate the uptake kinetics. Field-derived sampler-air 20 partitioning coefficients (K PE-A s) for OCPs were significantly correlated against the compounds’ 21 subcooled liquid vapour pressure (log P L ): [log K PE-A = -0.77+0.07 * log P L + 6.35+0.13 (R 2 = 22 0.90; n = 17; SE = 0.19; p < 0.001)]. Estimated and measured OCP concentrations in Alexandria 23 agreed well (factor difference ≤ 2) indicating the feasibility of monitoring OCPs using LDPEs. 24 OCP concentrations ranged from <LOD to 168 pg/m 3 . Calculated isomeric ratios indicated 25 recent usage of chlordanes and endosulfans.


Capsule Abstract
LDPE passive air samplers spiked with PRCs can be used as a practical low cost technique for monitoring gas phase OCPs in Egypt.
In some developing countries, however, several OCPs are still in use (Baek et al., 2013).
Pesticides were introduced in Egypt in 1952 and about one million metric tons of commercial pesticides were used until 2003 (Mansour, 2004).In addition, illegal pesticides application cannot be ignored due to the poor enforcement of environmental laws.In Egypt, 70 % of the pesticides are applied on cotton crops, while the rest are applied on corn, rice, sugarcane plantations, vegetables and fruits (Mansour, 2004).DDT, lindane and endrin were among the pesticides used to control cotton pests.Three decades ago, the use of DDT in agriculture was officially banned.Following the restriction on DDT, the use of other organochlorine pesticides (e.g., aldrin, dieldrin, chlordane, heptachlor, lindane) was gradually restricted in Egypt (Abou-Arab et al., 1995;Mansour, 2004).
Despite the fact that Egypt was a signatory of Stockholm Convention held on May, 2002, and the use of OCPs was banned in Egypt, these toxic compounds are still detected in various physical and biological environmental compartments in the country (EL Nemr and Abd Allah, 2004;Sallam and Morshedy, 2008;Barakat et al. 2012 a,b;Khairy et al., 2012;Barakat et al., 2013).To our knowledge, limited studies have been carried out to assess the atmospheric levels of OCPs in Egypt, possibly due to the difficulties associated with the conventional active Passive air samplers have been proven to be a powerful monitoring technique for POPs, as they are cheaper and require less labour compared to active samplers (Lohmann et al., 2001).
The purpose of the current study was to investigate the possibility of using LDPE as passive air samplers for OCPs in the atmospheric environment of Alexandria, Egypt.
Accordingly, we undertook two major 21-days sampling campaigns of 15 deployments during summer (July, 2010) and winter (January, 2011) across 11 different sites.By deploying samplers in different locations and in different seasons, we were able to assess the pollution levels, temporal variations and the possible sources of OCPs in Alexandria using isomeric ratios.

Site Description
Alexandria is the second-largest city of Egypt, with a total surface area of 2300 km and occurred mainly in the south and southeastern parts (Switch, 2011).

Preparation and Deployment of LDPEs
LDPE sheets were cut from commercial sheeting (Carlisle Plastics, Inc., Minneapolis, MN) with a thickness of 51 µm, yielding a 10 x 30 cm strip of ~1-2 g each.Samplers were cleaned with DCM and n-hexane and spiked with four performance reference compounds (PRCs) according to the method developed by Booij et al. (2002)
PUF samples were extracted using a Dionex ASE 350 (Dionex Corporation, Sunnyvale, CA 94088) accelerated solvent extraction device after spiking with the surrogate standard mixture.Extracts were concentrated to a final volume of ~1 mL using a rotary evaporator (after solvent exchange into hexane) and passed on a glass chromatographic cleanup column packed with silica gel/alumina (2:1 wt/wt) in order to remove the interfering compounds (Khairy and Lohmann, 2012).The collected fraction was concentrated to a final volume of ~ 50 µL.Finally, 2,4,6-tribromobiphenyl was added as the injection standard before analysis.

Instrumental Analysis and Quality Control
All  Table (SI 3) and text (SI 3).
Procedural blanks, field blanks, matrix spikes and duplicate samples (20 % of the total samples) were included with each sample batch, and were carried throughout the entire analytical procedure in a manner identical to the samples.OCPs were completely absent from procedural and field blanks indicating negligible contamination during transport, storage and analysis.
Limits of detection (LODs) were determined as the concentration of analytes in a sample giving a peak with a signal-to-noise (S/N) of 3 (Doong et al., 2002).(QA/QC procedures are detailed in text SI 4 and Table SI 3).

Physico-Chemical Properties
Internally consistent air-water partitioning coefficients (K AW ) and octanol-air partitioning coefficients (K OA ) were obtained from Schenker et al. (2005).Missing K OA values were obtained by correlating available K OA values from Schenker et al. (2005) against values obtained from EPI Suite (USEPA, 2011) (Figure SI 2).Missing K AW values were calculated according to equation 1 (Reinhard and Drefahl, 1999): where H c is Henry's law constant (atm.L.mol -1 ), R is the gas constant (0.08206 in L.atm/mol.K) and T is the absolute temperature in Kelvin.H c for δ-HCH was obtained from Meylan and Howard (2005).Values for oxychlordane, trans-nonachlor, o,p'-DDD, o,p'-DDT and endosulfan sulfate were obtained from SPARC online calculator (V4.5).Values for endrin aldehyde, endrin ketone and methoxychlor were obtained from Mackay et al. (2006).Internally consistent values of the sub-cooled liquid vapor pressure (P L /Pa) were obtained from Shen and Wania (2005).
Missing P L values were obtained by correlating values from Shen and Wania (2005) where T 2 and T 1 are the mean temperatures of the deployment and at which the partitioning properties were determined (K).

Uncertainty associated with K PE-A calculations
Calculated overall uncertainty combined the uncertainty of predicted partition coefficients, equilibrium adjustment of polyethylene concentrations based on the use of PRCs Deleted: and ∆Hvap is the enthalpy of vaporization (J mol -1 ) (the fraction equilibration remaining at time of PE sampler collection), and the uncertainty associated with the analysis of the OCPs (See text SI 6 for more details).K PE-A values for undetected OCPs (β-HCH, δ-HCH, dieldrin, endosulfan sulfate and methoxychlor) in the field study were predicted from a simple linear relationship with P L (see section 3.2, equation 5).
Accordingly, we used the same relative uncertainties (RU) in predicted K PE -A s as given for the log P L by Shen and Wania (2005).Since internally consistent P L values did not exist for δ-HCH, endosulfan sulfate and methoxychlor, they were assigned the highest uncertainty estimate of P L (500 %).The overall uncertainty (Table 1) ranged from + 43.0 % to + 504 % (0.20-0.80 log units), with higher uncertainty values (303-504 %) for predicted K PE-As compared to the field measured ones (43.0-65.0%) owing to the greater uncertainty estimates of their P L .
Table 1 3. Results and Discussion

Uptake of OCPs by the LDPE
The addition of PRCs to the samplers before deployment is useful to infer the exchange rate kinetics, assuming that the elimination rates (of PRCs) and uptake rates (of native compounds) are equivalent (Bartkow et al., 2004) at α = 0.05, p < 0.001).Moreover, loss rate constants for each of the PRCs were significantly higher in the winter compared to the summer season [Mann-Whitney rank sum test for PBB 9 and PBB 103: T = 66 and 73 respectively at p < 0.001; t-test for PBB 52 (t = -5.56 at p < 0.001) and OCN (t = -2.37 at p = 0.028)], which was probably related to the increased wind speed during the winter season (Table SI 1).This indicates that the chemical exchange between the LDPE sampler and the air was limited by the air side boundary layer resistance (k e α 1/K PE-A) based on the following relation: where a k is the mass transfer coefficient through the air side boundary layer and A s and V s are the sampler surface area and volume respectively.
A correlation between log transformed k e values and log K PE-A of the PRCs was established to extrapolate k e for all the investigated OCPs (Tables SI 6 and 7).At the end of the 21 days deployment period, the average values of % equilibrium for OCPs (Figure 1) in the samples indicated that OCPs with higher P L values were approaching equilibrium (a-HCH, g-HCH, HCBz and aldrin) (> 80 % equilibrium), whereas OCPs like DDTs were still far from reaching equilibrium (< 60 %).To gain more information about the accumulation patterns of OCPs, three LDPEs were deployed in each season at one of the sites (site 11), and one was harvested after each week of exposure period (Figure 1).The equilibration of αand γ-HCH, HCBz, aldrin, heptachlor, trans-and cis-chlordane after each week of deployment (week 2week 1 and week 3 -week 2) started to deviate from the linear uptake after three weeks deployment period in the summer season (Figure 1a), indicating that they were approaching

Sampler-air partitioning constants (K PE-A )
PE-A partitioning constants were calculated directly for 17 OCPs which were detected both in the passive sampler (corrected for non-equilibrium) and the PUFs at site 2 during the field deployments (Table 1).To predict values for the other investigated OCPs, we examined correlation between our measured K PE-A s and literature K OA values, which has been used previously to predict K PE-A s for PAHs (Bartkow et al., 2004;Khairy and Lohmann, 2012).As shown in Figure (SI 5), a highly significant and strong linear correlation can be observed for PAHs (R 2 > 0.90, p<0.001), whereas for OCPs, the relation explained only 32 % of the total variability in the data.
We observed a significant and strong correlation when K PE-A s were regressed against P L (Pa) (p < 0.001) (Figure 2a).The correlation explained 90.0 % of the total variability in the data.
The slope was insignificantly different from -1 (p < 0.001; R 2 = 0.90), implying that P L can be used as a good predictor of K PE-A, and that the magnitude of K PE-A for OCPs is dominated by their volatilities.The derived K PE-A were calculated from two 3-week field deployments and Deleted: Deleted: i constrained by our ability to safely operate active sampling equipment during the entire field work.Yet the strong correlation with P L implies that we were able to derive vales that are well constrained by physico-chemical properties, validating the field-derived data.Accordingly, K PE-A values for the non-quantified OCPs in the passive sampler were predicted (Table 2) based on the derived P L -K PE-A relation (equation 5): (5) (n = 17; R 2 = 0.90; SE = 0.19) Table 2 Figure 2 We propose that K PE-A s for the non-quantified OCPs in the field calibration study can be best predicted from equation ( 5) despite of the higher degree of uncertainty for endrin, endrin aldehyde, endosulfan sulfate, endrin ketone and methoxychlor (Table 1) as internally consistent physico-chemical properties were missing.For further assessment, calculated K PE-A values were compared with K PE-A values derived using sampler-water partitioning coefficients (K PE-W ) and air-water partitioning coefficients (K AW ) according to equation 6: 6) Calculated (from the field experiment) and predicted (K PE-W /K AW ) K PE-A values agreed well within 94 % (factor difference < 0.4 log units) for the majority of the investigated OCPs (Figure 2b).The good agreement observed for most OCPs supports the assumption that our fieldderived K PE-A values are good approximations of their real values, despite the limited deployment period in the field.Contrarily, lower degree of agreement was observed for endrin logK = -0.770.07 .logP + 6.35 0.13 PE-A L ± ± aldehyde, endrin ketone, endosulfan sulfate, trans-nonachlor and oxychlordane (factor difference ranged from 0.60-3.8log units).One explanation for this lower agreement is related to the degree of uncertainty associated with the partitioning coefficients (uncertainty from P L and K AW + uncertainty from the K PE-A and K PE-W partitioning coefficients).characterized by high traffic and industrial activities (site 1, 4, 6 and 9).The average HCBz concentration in Alexandria was only slightly higher than its northern hemisphere average concentration (~ 50 pg/m 3 ) (Jin et al., 2013), but much lower than measured concentrations in China (Liu et al., 2009) and Korea (Li et al., 2007) (Table SI 10).Although HCBz production has ceased in most countries, it is still being generated inadvertently as a by-product and/or impurity in several chemical processes, such as the manufacture of chlorinated solvents, chlorinated aromatics and pesticides.It is also released to the environment by incomplete combustion, and from old dumpsites (Barber et al., 2005).

Spatial distribution and seasonal variations of OCPs
Table 3 αand γ-HCH were the only detected HCH isomers in the atmospheric environment of Alexandria (Figure SI 7).In all the samples, α-HCH concentrations (<LOD -98.0 pg/m 3 ) were much higher than the γ-HCH (<LOD -36.0 pg/m 3 ), especially in the winter season.
Aldrin was detected in 82 % of the investigated PE samples with relatively high concentrations (<LOD -147 pg/m 3 ) compared to HCHs and HCBz (Figure SI 7).Chlordanes were detected in the majority of the investigated samples.Based on the average concentrations, trans-chlordane showed the highest concentrations in all the samples during both seasons (summer: 19 pg/m 3 ; winter: 9 pg/m 3 ) followed by cis-chlordane (8.0; 4.0 pg/m 3 ).Oxychlordane, the metabolite of chlordane, was only observed in the summer season with an average concentration of 7.0 pg/m 3 .Heptachlor epoxide, the metabolite of heptachlor was also observed only in the summer season (higher temperature).The detection of both metabolites only in the summer suggests that they were produced from the photodegradation of parent compounds (Park et al., 2011).

Deleted: Average
Deleted: is Deleted: in the northern hemisphere Endosulfan is still in use globally for agriculture and also for the control of disease vectors (Pozo et al., 2011).Concentrations of endosulfans ranged from 9.0-160 pg/m 3 and <LOD-65 pg/m 3 in the summer and winter seasons respectively.In all the samples, endosulfan I concentrations were higher than endosulfan II, and endosulfan sulfate was <LOD in all the samples.
p,p'-DDT was widely used in Egypt for agricultural purposes and for the control of disease vectors.Although officially banned in 1988, its was recently detected in sediments of the Egyptian Mediterranean Sea coast (Khairy et al., 2012) and in coastal lakes (Barakat et al., 2012a, b), because of its long environmental half-life time.In the current study, concentrations of HCBz concentrations in Alexandria were higher than concentrations measured in Europe but lower than in Asia (Table SI 10).Chlordane concentrations were among the highest when compared to concentrations in other locations worldwide using PUF passive samplers).γ-HCH and endosulfans on the other hand were lower than concentrations measured in parts of Europe, Canada, India, China and in the other African countries such as Ghana and South Africa.

Source identification of OCPs
Several isomeric ratios were applied (Figure SI 8) in the current study to determine the source of the atmospheric OCPs (HCHs), and whether OCPs were released to the atmosphere as a result of fresh application or an aged use (HCHs, chlordanes, endosulfans and DDTs).
In the current study, α-HCH/γ-HCH ratio ranged from 1.3-2.9(Figure SI 8) in the summer (hot) season, and 3.2-6.7 in the winter (cold) season indicating a possible seasonal revolatilization of old lindane residues.Lower ratio values observed in the summer season could be related to the increased lindane volatilization from soil as a result of elevated atmospheric temperatures.
Technical chlordane has been widely used as agriculture pesticide and termiticide in all over the world (Syed et al., 2013).Calculated ratios of trans/cis chlordane in the current study ranged from 1.5-2.9(Figure SI 8) in both seasons, indicating fresh application of technical chlordane (Hinkcley et al., 1990, Jantunen et al., 2000).According to the UNEP, the use of chlordane is severely restricted and limited to non-agricultural uses in Egypt (UNEP, 2003).
Endosulfan comprises two isomers, endosulfan I and II, with a I/II ratio of about 2.33 (Daly et al. 2007) sampling techniques.According to the Stockholm Convention, parties to the Convention should develop a national inventory of POPs.No information exists about the levels, sources and health effects associated with the exposure to atmospheric OCPs in Egypt.Yet frequent measurements of air concentrations in different locations as well as monitoring studies on various levels from local point sources up to the continental scale are a matter of a great importance.Accordingly, cheap techniques that accurately measure atmospheric concentrations of POPs are highly desirable.
2 and a population of 4.1 million, extending about 90 km along the coast of the Mediterranean Sea on the northwestern side of the Nile Delta (Figure SI 1).The city's climate shows Mediterranean characteristics, namely mild, variably rainy winters and hot summers that, at times, can be very humid.More details about the study area can be found in Khairy and Lohmann (2012).The land use map of Alexandria (Figure SI 1) indicated the existence of different land uses including the agriculture, which represented ~ 27% of the governorate area samples were analyzed for OCPs with an Agilent 6890 Series GC System connected to an Agilent 5973 Network Mass Selective Detector.DDTs ( p,p'-DDE, o,p'-DDD, p,p'-DDD, o,p'-DDT and p,p'-DDT), hexachlorocyclohexanes (HCHs: α-, β-, γand δ-HCH isomers), chlordanes (trans-and cis-chlordane isomers, oxychlordane and trans-nonachlor), heptachlor and its epoxide, aldrin, dieldrin, endosulfans (endosulfan I and II isomers and endosulfan sulfate), hexachlorobenzene (HCBz), endrin, endrin aldehyde, endrin ketone and methoxychlor were investigated.Separation of OCPs was accomplished with a DB-5 MS fused silica capillary column (30 m×0.25 mm i.d., 0.25 µm film thickness, J&W Scientific).Abbreviations of OCPs and more details on the instrumental analysis are given in equilibrium.Contrarily, endosulfan I and II, trans-nonachlor, endrin, oxychlordane, p,p'-DDE and p,p'-DDT increased linearly during the sampling period.During the winter season (Figure1b), only p,p'-DDD/o,p'-DDT, o,p'-DDD and p,p'-DDT increased linearly during the sampling period, whereas all the other detected OCPs were approaching equilibrium.When the approached % equilibrium results of OCPs based on the PRCs at site 11 were compared to the variation in the accumulated amounts of OCPs in the LDPE (ng/g PE) with time (Figure SI 6), good agreement was generally observed for the predicted stage of exchange of OCPs between the LDPE sampler and air in both seasons.This implies that PRCs can be used to calibrate sampler/site specific mass transfer behavior, and thus accurately estimate the gaseous concentrations of OCPs (and other POPs) in the atmosphere.
]: Didn't you already comment on HCB above?This last paragraph should either be deleted, or you paste the sentences in the relevant sections above.

3. 5 .
in the technical mixture.Endosulfan II is more reactive and so an elevated I/II ratio is representative of an aged signature, whereas ratios closer to the starting technical mixture suggest recent application.As shown in Figure(SI 8), calculated values in the current study were around the value of the technical mixture (2.3) indicating recent application of endosulfans.In this study, the ratio of p,p'-DDE/p,p'-DDT ranged from 2-106, indicating an aged application of DDT in the area.Comment [RL2]: Needs ref -maybe from Bidleman's work?? Evaluation of the Predictive Ability of the LDPE Samplers Estimated gaseous OCP concentrations calculated from the LDPE passive samplers corrected for non-equilibrium were compared with the gaseous concentrations actually measured in the atmosphere using the high volume sampler.Results are given in Tables(SI 8 and 9).Although the difference in the sampling periods between the passive (21 days) and active (2−3 days) sampling periods adds uncertainty to the results in Alexandria (Egypt), good agreement was generally observed between estimated and measured OCP concentrations in all the investigated samples (Figure3 a & b).During the summer and winter seasons, the difference ranged from a factor of 0.8 to 2.0.Additionally, slopes of both regression lines were not significantly different from one (at α = 0.05, P < 0.001) supporting this good agreement.The significant correlations obtained here suggest that the gas-phase OCP concentrations in Alexandria might not display significant short-term variations during our measurement periods.This assumption is supported by the observed statistical insignificant difference (p = 0.128) between the weekly measured gas-phase concentrations at site 2 (Figure SI 9) sampled with the conventional active sampling method.Overall, LDPEs have proven to be a powerful tool for predicting atmospheric OCP concentrations in the present study. Figure3

Estimated atmospheric concentrations based on LDPE PRCs
, impregnated in the polyethylene samplers before their deployment, were used to gauge whether OCPs had achieved equilibrium and to adjust for disequilibrium in polyethylene (C LDPE ) assuming that uptake and elimination rates are equivalent.(Booijet al., 2002) (Text SI Lohmann (2012)ner (2002)Mackay et al. (2006)(Figure SI 3).P L value for trans-nonachlor was obtained fromShoeib and Harner (2002).Values for endrin aldehyde, endosulfan sulfate and endrin ketone were obtained from EPI Suite.Enthalpies of vaporization (∆H vap in KJ/mol) were obtained fromShoeib and Harner (2002).Missing ∆H vap values were obtained by correlating available values against P L (FigureSI 4).Sampler-water partitioning coefficients (K PE-W ) were calculated according toLohmann (2012).Values of all the physico-chemical parameters are given in Table (SI 4).(day -1 ), t is the deployment period (days) and K PE-A is the sampler-air partitioning coefficient (L/kg).To estimate K PE-A values for OCPs, a high volume sampler was left operating for a period of 15 days (5 days per week) at one of the sampling sites (site 2) during the summer season (See text SI 2 for more details).DetectedOCP concentrations in the LDPE and PUF samples were used to calculate the K PE-A values according to equation 2. Partition coefficients were adjusted for temperature according to a modified form ofVan't Hoff equation (equation 3) Wallis one way ANOVA statistical test indicated that loss rate constants decrease with increasing the sampler-air partitioning coefficient (log K PE-A ) (PBB 103 < OCN < PBB 52 < PBB 9) in both seasons (H summer = 39.1,H winter = 38.6 . Exchange kinetics were quantified by k e (for details, see textSI 5).Four PRCs with different log K PE-A values were used in this study.The least dissipated PRC in all the investigated samples was pentabromobiphenyl (PBB 103) (TableSI 5).A higher elimination rate was observed for PBB 103 during the winter (30-46 %) compared to the summer (25-33 %).The Kruskal-