Organochlorine pesticides in ambient air in Durban, South Africa

S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r. com/ loca te / sc i to tenv

Organochlorine pesticides in ambient air in Durban,South Africa

S.A. Battermana,⁎, S.M. Chernyaka, Y. Goundenb, M. Matooaneb, R.N. Naidoob

aEnvironmental Health Sciences, University of Michigan School of Public Health, 109 Observatory St., Ann Arbor, MI 48109-2029, USAbCentre for Occupational and Environmental Health, University of KwaZulu-Natal, 719 Umbilo Road, Private Bag 7,Congella, 4013 Durban, South Africa

A R T I C L E I N F O

⁎ Corresponding author. Tel.: +1 734 763 2417E-mail address: stuartb@umich.edu (S.A. B

0048-9697/$ – see front matter © 2008 Elsevidoi:10.1016/j.scitotenv.2008.02.033

A B S T R A C T

Article history:Received 21 September 2007Received in revised form4 February 2008Accepted 14 February 2008Available online 8 April 2008

Despite the existence of numerous sources and continuing use, information regardingemissions and airborne concentrations of organochlorine pesticides in Africa is extremelylimited. This paper presents results of a monitoring program conducted in Durban, SouthAfrica that was designed to characterize levels, trends and possible sources of pesticides inboth industrial and residential areas. Three monitoring sites were established, two in anindustrialized area in the southern part of the city, and the third in a northern residentialarea. Particulate and vapor samples were sampled over the 2004–5 period and analyzed byGC/MS to estimate long-term levels of awide range of pesticides. Based on ayear of sampling,the sites had comparable levels of many pesticides with exceptions of α-chlordane andlindane. Levels of p,p'-DDT (42±27 pg m−3) and its derivatives were relatively high andshowed an unusual mixture with high levels of p,p'-DDD (12±11 pg m−3). Other pesticidesdetected and quantified included aldrin, chlordanes, hexachlorobenzene and dieldrin.Potential source areas, identified using concentration patterns, local and regional gradients,compositional information and trajectory analyses, suggest that chlordane and lindane arisefrom both local sources as well as regional/global sources; DDT from regional sourceselsewhere in South Africa, Africa and India; and most of the other long-lived pesticidesdetected, including γ-nonachlor, hexachlorobenzene and toxaphene, from global sources.This monitoring results, which represent the most detailed study to date of pesticidesin air in Africa, serve several purposes, including documenting the presence and useof long-banned pesticides like aldrin, aiding the understanding of the fate of persistentcompounds, identifying pollutants that may contribute to health problems, and informingdecision-making aimed at reducing exposures and risks.

© 2008 Elsevier B.V. All rights reserved.

Keywords:Airborne contaminantsAfricaDDTLindaneOrganochlorinesPesticidesTransport

1. Introduction

Atmospheric transport is a primary route for transportingpersistent organic pollutants (POPs) such as polychlorinateddioxins, PCBs, and DDTs (Wania and Mackay, 1993; Bawdenet al., 2004; Shen et al., 2005). POPs can be emitted by manysources, including pesticide applications, uncontrolled com-bustion (including the open waste burning that is prevalent

; fax: +1 734 763 8095.atterman).

er B.V. All rights reserved

throughout much of Africa), and volatilization from residuespresent in soils. Despite the existence of numerous sourcesand continuing use, emissions of POPs in Africa are, at best,poorly documented, and airborne monitoring is extremelylimited. Most African countries, including South Africa, havesigned and ratified the Stockholm Convention on PersistentOrganic Pollutants (POPs), and therefore are presumablyphasing out the use of most organochlorine pesticides. It is

.

120 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

notable that lindane is not in the Convention's POPs list, andthat DDT uses for malaria control are exempted.

This paper presents results from an ambient air qualitymonitoring program conducted in Durban (eThekwini Muni-cipality), South Africa, a large metropolitan area located onAfrica's southeast coast. This area has had air quality con-cerns for many years. Most attention has focused on localizedproblems in the Durban South Industrial Basin (DSIB), adensely populated area with intense industrial activity andproblems related to SO2, odors, H2S, HF, and flaring at therefineries (Reid and D'Sa, 2005). There are no known previousmeasurements or estimates of airborne emissions or concen-trations of pesticides (or other POPs) in the region. Establishedfollowing a stakeholder process coordinated by the Municipa-lity's health authority, a monitoring program was initiatedthat included measurements of both conventional and toxicpollutants throughout the area. The objectives of this paper

Fig. 1 –Maps showing locations of monitoring sites in greater Durin the DSIB. Inset maps show province of KwaZulu-Natal and So

are to characterize levels and trends of pesticides at twolocations in the SDIB and at a comparison site approximately20 km distant, to compare concentrations in Durban withlevels measured elsewhere, and to identify the presence oflocal sources contributing to airborne levels. We examine anumber of older (legacy) persistent pesticides and relatedorganochlorinated POPs, i.e., DDT, lindane, pentachloroben-zene, octachlorostyrene, and pentachloroanisole.

2. Background

2.1. Geography and land use

TheDSIB is located on the east coast of SouthAfrica (Fig. 1) andis part of a shallow basin comprising a ~4×24 km coastal strip.This flat alluvial corridor is defined to the NE by the inland

banmetropolitan area. The two southern monitoring sites lieuth Africa.

121S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

Berea Ridge and Southern Freeway, to the SW by the BluffRidge, and to the SE by the Indian Ocean. The two ridges rise70–110m aboveMSL and run parallel to the coastline (SW–NE).Several rivers, including the Umhlatuzana, Umbilo andUmgeni, flow from highlands SSE into the Indian Ocean.Inland, elevations rise gradually and then sharply to thesubcontinent's plateau. At a distance of ~120 km inland,the 3000+ m Drakensberg Mountains form the border be-tween the province of KwaZulu-Natal (KZN) and the country ofLesotho.

Laying just south of Durban's central business district, theDSIB is the largest urban and industrial area in KZN, with~200,000 people living in 25 designated “suburbs,” most ofwhich remain racially segregated. Residential and industrialareas are intermingled. The area has one of the highestconcentrations of industrial activity in Africa, containing twolarge petroleum refineries, a paper mill, an internationalairport, a large chemical tank farm, landfill sites, incinerators,numerous processing and manufacturing industries, majortrucking, harbor and rail facilities, and other industries. Thisarea was believed to be at high risk of experiencing highpollutant concentrations due to the density of large industrialand fugitive emission sources, the use of short exhaust stacksat these facilities, the lack of effective emission controls, andthe topography and coastal environment that promotesrecirculation of air.

2.2. Meteorology, and air quality

Concentrations of conventional pollutants, e.g., SO2 and PM10,in Durban show strong seasonal patterns that can beattributed to the frequency and depth of inversions, alongwith other factors. In summer, macroscale low pressuresystems prevail over South Africa's interior, unstable condi-tions are frequent, and the resulting convective stormsproduce strong updrafts and downward flows of fresher andcooler air (Preston-Whyte and Tyson, 1989) that promotepollutant dispersal and removal. In winter, the high-pressurecells shift northwards, and persistent highs are common,producing lightwinds and occasionally unfavorable dispersiveconditions. These conditions are frequently (five to six timesper month) interrupted by traveling low pressure systemsassociated with the development of frontal systems andmoderate winds that ameliorate pollutant concentrations. Intheir prefrontal stage, the shallow coastal low promotessubsidence, inversions and light katabatic mountain-plain“Berg” (Drakensberg mountain range) winds in which cool airfrom the mountains drains towards the coast. These condi-tions inhibit mixing and also transport pollutants from theinterior to the coast, and thus they are associated with calm,warm and stable conditions that may last several days anddeclining air quality (eThekwini, 2004). Winter-time tempera-ture inversions are common, and may be strengthened bykatabatic flows (Simpson and McGee, 1994). By late morning,nocturnal inversions dissipate, but fumigation effects (whenelevated plumes intersect a growing thermal boundary layer)may produce high ground-level concentrations. The passageof the front is often associated with rain, the break-up of theinversion, and a deep mixing layer which promotes mixingand pollutant washout, and which lowers pollutant levels.

Prevailing winds are from the NE and SW. SW winds aregenerally stronger and may be accompanied by rain. Windsare influenced by the local and regional geography, includingthe large scale “Berg” wind system (mentioned above) whichcombines with cool winds from the Umhlatuzana, Umbilo andthe Umgeni River mouths and the offshore land breeze (Tysonand Preston-Whyte, 1972; Wekker et al., 1998; Preston-Whyteand Tyson, 1989). The Bluff and the Berea Ridges help steerthese winds to the SW. The temperature inversion associatedwith these low speed wind systems produces stable atmo-spheres, reduced vertical mixing, and fanning plumes extend-ing to the sea. These shallowwinds have historically producedhigh SO2 concentrations in low lying coastal areas, but lowconcentrations at elevated areas, e.g., the Berea ridge. Thewinds from the Umgeni River blow to the sea and, in thedaytime, may become entrained by the NE gradient winds andsea breeze and brought back to the city (Preston-Whyte andTyson, 1989). Because they are associated with unstableatmospheric conditions, sea breezes favor pollutant dispersaland lower concentrations.

3. Materials and methods

3.1. Monitoring sites

Monitoring was conducted at three sites (Fig. 1). In thesouthern part of the DSIB, the Nizam site was established ata primary school located in a small residential area amidstmajor industrial activities: the Engen petroleum refinery(0.73 km NE); the Southern Works waste water treatmentplant (0.81 km NW); the Mondi pulp paper plant (0.85 km W);and the Sapref refinery (0.88 km SW). The first monitoring sitewas on a paved surface and close to a grassy bank and ~10 mfrom the school building. The second site, Wentworth, waslocated in the central portion of theDSIB. This site is located ona ~1.5 ha grassy knoll, adjacent to a wooded area to the E,residential areas to the S and W, many smaller industriesfurther (~0.5 km) to theW, a large hospital anda cemetery to theN. The third site, Ferndale, was located ~20 km north of theDSIB,well on theother sideofDurban's central business district.This site was located on a grassy patch at a primary school ona hilltop in moderately hilly terrain, in a residential with asignificant amount of open space. The closest industry is thenew River Horse Valley and Phoenix industrial parks (~4 kmto the NE).

3.2. Sampling

Monitoring was conducted from August 2004 to September2005. The schedule called for simultaneous measurements atthe three sites every 12 days over a one year period, whichwould have resulted in 29 sampling events. Here we report on13 to 17 valid sampling events at each site, a somewhatreduced number due to sampler failures, power outages,supply and shipping logistics, and other problems.

New high volume air samplers (TE-1000, Tisch Environ-mental) were installed at each site. Particulate matter wascaptured on quartz microfiber filters (102 mm dia, WhatmanInternational Ltd, QMA-4 filters), and vapor phase pollutants

122 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

were collected on polyurethane foam (PUF) plugs (75 mm dia,SKC). All sampling and shipping materials were cleaned andchecked prior to deployment, e.g., quartz filters were baked for5 h at 400 °C, and PUF plugswere placed in a Soxhlet apparatusand precleanedwith hexane for 8 h. All shipping and transportmaterials were precleaned using hexane rinses and baking atthe University of Michigan. All sensitivematerials were sealedand checked for possible contaminations following shipping,and all blanks met QA/QC criteria. In the field, filters and PUFpugs were spiked with surrogates, e.g., PCB-65 and PCB-166immediately after sampling, to determine any matrix effectsand recoveries. Flows rates were checked before and aftersampling following manufacturer's instructions and a stan-dard protocol, and averaged 230 L min−1. Sampling periodswere 24 h (a few longer samples were taken) and the averagesample volume was 315 m3.

Conventional pollutants were measured during the studyperiod, and we contrast trajectories of pesticides with PM10

levels averaged across four Durban monitoring sites. We alsoobtained surface and upper air meteorological data collectedat nine sites in the area, including those operated by theMunicipality, airports, and others.

3.3. Sample analysis

After collection, samples were shipped to our University ofMichigan laboratory and refrigerated until analysis. Filters andPUF plugs were Soxhlet extracted for 36 h per sample usingbenzene/hexane (4:1) for a multiple analyses, including poly-chlorinated dioxins and furans, PCBs, PAHs and pesticides. In

Table 1 – Pesticide concentrations (pg m−3) averaged across thr

Compound Particulate phase Vap

Average St.Dev. Max n Average St

Pentachlorobenzene 0.15 0.12 0.66 48 3.50α-HCH 0.09 0.15 0.62 36 1.50Hexachlorobenzene 0.10 0.12 0.41 42 4.50Pentachloroanizole 0.06 0.07 0.23 33 14.00Lindane 8.70 8.60 39.00 45 124.00 1Aldrin 0.00 0.01 0.06 1 10.10Dachtal 0.01 0.01 0.05 12 0.40Octachlorostyrene 0.00 0.01 0.04 6 0.01Heptaclorepoxide 0.05 0.09 0.37 19 0.58Oxychlordane 0.01 0.02 0.07 15 0.27γ-Chlordane 0.17 0.27 1.21 46 9.30α-Chlordane 0.04 0.02 0.09 46 10.80γ-Nonachlor 0.05 0.04 0.21 46 3.80Dieldrin 0.00 0.01 0.04 4 0.00p,p'-DDE 3.90 4.10 19.00 46 8.50Endrin 0.06 0.31 1.93 4 0.00α-Nonachlor 0.04 0.03 0.14 39 1.00p,p'-DDD 11.00 7.30 30.00 48 14.10p,p'-DDT 15.00 11.70 48.00 47 27.00Hexa-Toxaphene 0.08 0.09 0.28 13 3.50Hepta-Toxaphene 0.09 0.09 0.24 14 5.80Octa-Toxaphene 0.01 0.02 0.05 7 1.40Nona-Toxaphene 0.01 0.01 0.06 7 0.65Deca-Toxaphene 0.02 0.09 0.37 2 0.47Total Toxaphene 0.28 0.41 1.64 14 11.00

Notes: n=number of detections. Fraction of concentration in vapor phase

brief, extracts were transferred into an evaporation flask,reduced to ~10ml (~20min each sample) in a Kuderna-Danishevaporator, and cleaned with 4.5 M sulfuric acid. The extractswere evaporated under N2 flow (2–3 h per sample), andtransferred into a GC vial. The final volume was typically1.0 mL. An internal standard was added to each sample.Separate sample preparation and GC/MS runs were performedfor particulate and vapor phases.

Four GC/MS runs were used to analyze each sample, twoeach for particulate and vapor fractions. Analyses wereperformed using aGC/MS (Model 6890/5973, Agilent Industries,Palo Alto, CA, USA), fused silica DB-5 column (30 m length,0.25mm id, film thickness 0.25 um, J&W Scientific, FolsomCA,USA), injector temperature of 280 °C, helium carrier gas witha flow of ~0.7 ml min−1, and source temperature of 150 °C.The MSD was operated in selected ion monitoring (SIM) modeusing negative chemical ionization mode and methane asthe reagent gas (Chernyak et al., 2005). Oven temperatureprograms and other parameters that varied by GC/MS run aredescribed below.

The first GC/MS run analyzed organochlorine pesticidesand other halogenated chemicals including (MDLs or met-hod detection limits in parentheses): aldrin (0.02 pg m−3); α-chlordane (0.03 pgm−3); γ-chlordane (0.03 pgm−3); α-nonachlor(0.02 pg m−3); γ-nonachlor (0.03 pg m−3); dacthal (0.02 pg m−3);dieldrin (0.01 pg m−3); endrin (0.05 pg m−3); hexachlorobenzene(0.01 pg m−3); lindane (0.05 pg m−3); mirex (0.02 pg m−3); octa-chlorostyrene (0.06 pg m−3); oxychlordane (0.02 pg m−3); p;p'-DDD (8 pgm−3); p;p'-DDE (3 pgm−3); p;p'-DDT (6 pgm−3); penta-chloroanisole (0.01 pg m−3); pentachlorobenzene (0.02 pg m−3);

ee sites

or phase Total Vaporfraction

(%).Dev. Max n Average St.Dev. Max n

2.10 8.70 48 3.60 2.10 9.00 48 95.91.00 4.10 48 1.60 1.10 4.20 48 94.22.40 8.90 47 4.50 2.50 9.00 47 100.0

11.00 45.00 48 14.00 11.00 45.00 48 99.635.00 472.00 48 133.00 141.00 504.00 48 93.514.20 49.00 30 10.10 14.20 49.00 30 100.00.50 2.50 39 0.40 0.50 2.50 39 98.50.04 0.23 9 0.02 0.04 0.23 12 81.30.75 3.65 39 0.62 0.80 3.70 39 92.30.20 0.72 48 0.28 0.21 0.77 48 96.76.60 27.00 48 9.50 6.60 28.00 48 98.29.50 35.00 48 10.90 9.50 35.00 48 99.61.90 9.80 47 3.90 2.00 9.90 48 98.70.01 0.05 3 0.00 0.01 0.06 5 46.37.30 30.00 48 12.30 10.60 45.00 48 68.80.00 0.03 2 0.06 0.31 1.93 4 1.61.00 4.00 48 1.00 1.00 4.00 48 97.27.20 30.00 48 25.00 11.10 53.00 48 56.6

18.00 72.00 48 42.00 27.00 120.00 48 63.31.00 5.20 18 3.60 1.00 5.30 18 97.92.20 10.00 18 5.90 2.20 10.00 18 98.40.90 3.10 18 1.40 0.90 3.20 18 99.30.54 1.90 18 0.66 0.54 1.90 18 98.70.61 2.20 14 0.49 0.63 2.20 14 95.64.00 23.00 18 12.00 4.00 23.00 18 97.6

(%) is at far right.

123S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

photo-mirex (0.02 pg m−3); α-HCH (0.05 pg m−3); and hepta-chlorepoxide (0.04 pg m−3). The oven temperature programstarted at 80 °C, held for 1 min, ramped at 20 °Cmin−1 to 150 °C,ramped at 2 °C min to 250 °C, held for 4 min, ramped at 10 °Cmin−1 to 300 °C, and held for 6 min. Mixtures of neat pesticideswere prepared, and three point calibrations were performed.

The second run analyzed five homologues of toxaphene(hexa, hepta, octa, nona, deca) fromwhich the total toxaphenewas determined. Here, the GC oven temperature programstarted at 80 °C, ramped by 10 °Cmin−1 to 220 °C, then 1.5 °C to230 °C, and finally at 10 °C min−1 to 300 °C and held for 6 min.Toxaphene standards used neat technical toxaphene pre-pared at concentrations of 500, 2500 and 5000 ng/mL of totaltoxaphene. The MDL for total toxaphene was 0.2 pg m−3.

3.4. Quality assurance

Quality assurance activities included routine use of blanks,spike recovery, and surrogate recovery determinations; co-location studies; and inter-laboratory comparisons of splitsamples and standards. Blanks were run with each samplebatch, separately for filters and PUFs. All solvents and othermaterials contacting samples were clean, as confirmed usingblanks. In all cases, blanks were clean for pesticides. Spikerecovery tests were performed during method developmentand periodically during analyses. In the latter, randomly

Fig. 2 –Relative changes (percent from mean) in concentrations foareas, as determined by back trajectory analyses. “4 Chlordanesp,p'-DDE exceed 50%.

selected PUFs and filters were spiked (using a mixture of allanalyzed pesticides) for recovery tests. All pesticides (includ-ing dacthal and dieldrin) showed acceptable recoveries (80 to101%). Surrogate recovery tests using unlabeled PCB IUPACStandards 65 and 166, which were performed on each sample,were acceptable (75 to 110%). Co-location tests using twosamplers were performed at one site (Wentworth), and resultsshowed good agreement, e.g., average of 8% difference forpesticides.

Analytical uncertainties were calculated from replicatelaboratory measurements, field duplicates, and analysis ofreference materials. This information was used to assessoverall measurement uncertainty. The MDLs noted abovewere at 99% confidence, and were estimated as 3 times thestandard deviation of 7 replicates of a procedural blank spikedwith a very low level of analyte (Keith, 1991).

3.5. Data analysis

For each contaminant, we calculated descriptive statistics bysite and phase (vapor or particulate), and also for data pooledacross sites. Trend plots were made for key pollutants, and the3-site average was computed. To show long-term trends, theaverage was exponentially smoothed (smoothing parameterα=0.01 day−1), followed by polynomial curve-fitting. Smoothedaverages were plotted along with observations. Typically, POPs

r six pesticides and particulate matter from seven source” is sum of chlordanes. Change for last three source areas for

124 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

are reported with particulate and vapor phases combined,however, we present phase-specific concentrations to identifysources and site differences. For DDT, we computed the DDE/DDT ratios and report the median and interquartile range. Thetotal chlordane concentration (Σ4chlordane) was calculated asthe sum of γ- and α-chlordanes and γ- and α-nonachlors. Giventhe many toxaphene components, analyses are presented ashexa, hepta, octa, nona, and deca chlorinated homologues, anda summary measure (total toxaphene).

Levels monitored in Durban were compared to pesticidelevels elsewhere. We focused on measurements in or nearAfrica, but since such monitoring is so scarce, we utilizedother data, including that collected by the CooperativeProgramme for Monitoring and Evaluation of the Long-rangeTransmission of Air Pollutants in Europe (EMEP; Aas andBreivik, 2005), which mainly uses six remote and northernEuropean sites to measure several pesticides with a sufficientnumber of samples to estimate annual averages. We con-solidated the annual averages from the three most recentyears (2002–4) for the available sites, and calculated mediansand other statistics for comparison to the Durban data. Wealso discuss POPs measured in other studies, including therecent Global Atmospheric Passive Sampling (GAPS) study,which mostly used background sites, including those inBotswana, Ghana and the Northern Cape, South Africa, atwhich a single seasonal sample was collected from Dec. 2004through March 2005 (Pozo et al., 2006).

Back trajectory modeling used the Hysplit model (Draxlerand Rolph, 2003) and the final run analysis (FNL) meteorolo-gical dataset to simulate 3-day back trajectories every 4 hduring each of the 24-h sampling periods, i.e., six backtrajectories were computed for each 24-h sampling event.The trajectories were then used to determine the number ofhours that each trajectory was over five large continentalareas (N KZN, S KZN, N interior Africa, S interior Africa) andthree ocean areas (N and S Indian Ocean, S Atlantic), whichwere then used as weights to obtain area-weighted concen-trations from each possible source area. The Nizam measure-ments, which had the most samples (n=19 in most cases),were used for this analysis. Differences in the area-weightedconcentrations can help to identify emissions that are local or

Fig. 3 –Comparison of average concentrations of selected pesticidbetween sites.

regional in origin, though the identification of source areas isrelatively crude given the duration of the trajectories, thelimited number of observations, the relatively coarse resolu-tion of the meteorological data (191 km), and other uncertain-ties. Given the small sample size, this analysis was notattempted for toxaphene.

4. Results and discussion

Table 1 summarizes concentrations of pesticides averagedacross the three sites. Concentrations in the vapor phase weremuch larger than particulate phase levels, with exceptions ofdieldrin and endrin. Trajectory analysis results for pesticides(and PM10 as a comparison) are summarized in Fig. 2, whilelevels of selected pesticides across the three monitoring sitesare shown in Fig. 3. Trends of four pesticides over the year-longmonitoring period are shown in Fig. 4, while levels of ninepesticides (or derivative products) over Europe (EMEP) andDurban are compared in Fig. 5.

4.1. Hexachlorobenzene (HCB)

The three sites showed similar concentrations of HCB (4.7±2.7 pg m−3 at Nizam, 5.3±2.8 pgm−3 atWentworth, 3.6±1.6 pgm−3 at Ferndale), suggesting long-range transport. The trajec-tory analysis (Fig. 2B) showed similar concentrations arrivingfrom each potential source area (within 7% of the mean exceptfor S KZN and S Atlantic), also consistent with global sources.

HCB is a fungicide introduced in 1945 for seed treatment,particularly against bunt which affects cereals such as wheatand rye (Barber et al., 2005). It is also a by-product of themanufacture of industrial chemicals, and a known impurity inseveral pesticide formulations (ATSDR, 2002). In the US, itsregistration as a fungicide was cancelled in 1984. In SouthAfrica, HCB was banned in 1983 (Vermeulen et al., 1998).Historically, the highest airborne levels have been found in thenorthern hemisphere where HCB was used or produced inlarge amounts, and very high levels were measured in the late1980s and early 1990s in the US, Canada, Germany, Spain,France and England (ATSDR, 2002; Jones, 2005), especially at

es at the three sites, arranged in order of increasing variation

Fig. 4 –Trends of selected pesticides at the three monitoring sites.

125S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

contaminated sites (Popp et al., 2000). The recent EMEP data(Aas and Breivik, 2005) show 55 pg m−3 (median at 3 sites,Fig. 5). The most recent southern hemisphere measurements

Fig. 5 –Comparison of Durban and 2002–4 EMEP data. Plotsshow 25th, 50th, and 75th percentile concentrations forDurban and EMEP data, plus minimum and maximum forEMEP data. EMEP sample size is 15–20 and uses annualaverages, except for HCB (n=10) and each chlordanes (n=6).

are 4–34 (average of 11±7.5) pg m−3 at Lake Malawi in 1997–8(Karlsson et al., 2000), average of 25–63 pg m−3 at 10 NewZealand sites in 1996–9 (Buckland et al., 1999), and 6–22(average of 11) pg m–3 in the Southern Ocean in 2001 (Jawardet al., 2004). All of these earlier measurements exceed Durbanlevels, suggesting a globally declining trend of this widelydistributed POP. We do not believe that our results reflectsignificant losses ofHCBsdue to breakthrough through the PUFcartridge, based on Popp et al. (2000) who used similar samplevolumes.

4.2. Chlordanes

Low to moderate concentrations of γ- and α-chlordane, andγ-nonachlor were detected, but levels of γ- and α-chlordaneincreased strongly from north-to-south (Figs. 3, 4B), e.g.,Σ4chlordane concentrations averaged 10±4, 24±5 and 38±18 pg m−3 at Ferndale, Wentworth, and Nizam respectively.Negligible site-to-site differences throughout the year wereseen for γ-nonachlor (Fig. 4C), and the highest concentrationsof both γ- and α-nonachlor were found at Wentworth. Asummer peak is seen for these compounds (Fig. 4B and C).

126 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

Chlordane source areas were similar to lindane, i.e., thehighest concentrations arrived from N central Africa and NIndian Ocean (Fig. 2C). Also, the high γ-chlordane/α-chlordaneratio (1.2 overall) is close to that of the technical mixture (1.56),indicating regional and relatively recent emissions of thispesticide since γ-chlordane's greater reactivity results in lowratios for aged materials (Pozo et al., 2006).

Chlordane was introduced as an agricultural pesticide in1946 and was widely used as a broad-spectrum contactinsecticide through the 1960s on vegetables, small grains,maize, other oilseeds, potatoes, sugarcane, sugar beets, fruits,nuts, cotton, and jute. It has also been extensively used tocontrol termites (Fisher, 1999). In the US, restrictions wereinitiated in 1974, sales were halted in 1988, and the majorproducer (Velsicol Chemical Co.) voluntarily halted globalproduction in 1997 (Offenberg et al., 2004). Technical chlordaneis a mixture of ~140 compounds, but most consists of α- andγ-chlordane (38–48%), heptachlor (3–13%), α- and γ-nonachlor(5–11%), and other chlordane isomers (17–25%). Chlordane ismetabolized to primarily oxychlordane. In South Africa, chlor-dane, dieldrin and other pesticides were imported through theport of Durban, and it is believed that local industriesreformulated, packaged and distributed these products. Chlor-dane usage in South Africa was restricted in 1993 to stemtreatment (e.g., vineyards) andbuilding/house treatment; itsusein agriculture was withdrawn in 2000 (South African Depart-ment of Agriculture, 2008). Durban area emissions appear likely,as suggested by this history and the pollutant gradient observedin Durban. Across the US and Canada, a latitudinal gradient ofΣ4chlordane in air has been observed, with average concentra-tions ranging from 1.5 pgm−3 (Alert, Canada) to 616 pgm−3 (LosAngeles, CA; Offenberg et al., 2004). The EMEP results fall below1pgm−3, thoughα-nonachlor is not included (Fig. 5). Both γ- andα-chlordane and γ-nonachlor can persist through decades ofweathering in soils (Incorvia-Mattina et al., 1999, 2002,). AtLake Malawi (1997–8), airborne levels of Σ4chlordane averaged51±68 pg m−3 (Karlsson et al., 2000). The GAPS measurementsin Africa showed low concentrations or non-detects for α- andγ-chlordane and γ-nonaclor.

Chlordane levels in Durban are comparable to those mea-sured in northern Michigan and New York, ~10-fold higherthan EMEP levels, and about half that seen at Lake Malawi.While the gradient of γ- and α-chlordane among the threesites indicates local applications to the south of Durban, thetrajectory analysis, low concentrations and lack of high con-centration peaks also suggest that local emissions are limitedor at an intermediate distance. The relatively low levels alsoreflect the general global decrease in concentrations of thisPOP.

Oxychlordane levels were very low in Durban (0.18 to0.42 pg m−3) compared to recent US measurements, butsimilar to levels at Lake Malawi (0.44±1.01 pg m−3; Karlssonet al., 2000). The oxychlordane/Σ4chlordane ratio is similar torecent US measurements (Offenberg et al., 2004).

4.3. Heptachlor and derivates

Heptachlorepoxide was found in most (vapor) samples at lowlevels, and averaged 0.62±0.80 pg m−3 across the three sites.The highest levels were found at Wentworth, with several

measurements in the 2–4 pg m−3 range. This non-systemicstomach and contact insecticide was used primarily againstsoil insects and termites, as well as cotton insects, grass-hoppers, some crop pests, and mosquitoes (Fisher 1999).Karlsson et al. (2000) reported heptachlorepoxide levels of0.82±1.9 pg m−3 at Lake Malawi, just slightly higher than ourmeasurements.

4.4. DDT and derivatives

The three sites showed generally comparable concentrationsof DDT-related compounds p,p'-DDT, p,p'-DDE, and p,p'-DDD(Fig. 3) and similar trends (Fig. 4D). The seasonality andsummer timemaximamay be associated with usage patterns,increased volatilization from soil at higher temperatures, andtransport. However, several factors indicate both recent anddistant uses of DDT. First, as discussed below, ratios of DDE/DDT concentrations are low (median=0.29, interquartilerange=0.62, n=48) and show seasonality (lower ratios insummer). Second, the spatial variability of DDT concentra-tions is limited, e.g., p,p'-DDT levels at Wentworth averaged50±22 pg m−3, 40±36 pg m−3 at Ferndale, and 37±21 pg m−3 atNizam (COV of only 15%). Third, peak levels exceeded averagelevels by only a factor of 2 to 3. Fourth, air masses arrivingfrom S KZN had elevated levels of p,p'-DDT (36% above themean) and still higher increases of p,p'-DDE levels (61%;Fig. 2D, E). Source areas in S KZN and S central Africa alsoshowed large increases in DDE (53% above the mean; Fig. 2E).One of the known current DDT use areas, the northern portionof KZN, was not associated with higher levels, possibly due touncertainties in trajectory analysis, e.g., insufficient length ofthe trajectories. Finally, while DDT and DDD trends weresimilar (r=0.77 at Nizam, n=19), DDT and DDE levels hadnegligible correlation (r=−0.20 at Nizam, n=19). The elevatedlevels of p,p'-DDD could be due to the use of dicofol (US EPA,1998), a miticide which includes several DDT analogs asmanufacturing impurities. Other derivatives of DDT wereevident in the chromatograms but not quantified. (Our calibra-tion standards included only parent compounds.)

DDT was used extensively for insect control during WorldWar II and subsequently as a pesticide on agricultural crops,especially cotton. While most agricultural uses have ceased,DDT is still used againstmosquitoes formalaria control. SouthAfrica banned DDT uses in 1983 except for malaria control,and it is used extensively in northern KZN where malaria isendemic and spreading southwards (Bouwman et al., 2006).High airborne concentrations of DDT have been reported inusage areas, e.g., 1,050 pgm−3 (n=10, vapor phase only) at LakeKariba, Zimbabwe in March 1994 (Larsson et al., 1995). Morerecent (1997–8) monitoring at Lake Malawi showed averagep,p'-DDT levels of only 12±27 pg m−3 (Karlsson et al., 2000).Durban levels were higher, e.g., the three-site p,p'-DDT con-centration averaged 42±27 pg m−3. With the exception ofmeasurements in the Czech Republic (8–16 pg m−3 at Kosice),the Durban measurements exceed EMEP levels by N100-fold,but of course DDT has not been used in Europe for decades(Fig. 5).

DDE and DDD are persistent dechlorination products ofDDT. High DDE/DDT ratios have long been used to indicateolder usage of technical DDT, especially in soil, sediment and

127S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

biological samples, and a DDE/DDT ratio of ~3 is normallyexpected for aged environmental mixtures (Strandberg et al.,1998). Technical DDT used in South Africa contains 72–74% ofp,p'-DDT (Bouwman et al., 2006). Less guidance is available forratios of airborne concentrations. In Canada and theUS, whereDDT has been banned for decades, the median DDE/DDT ratioof airborne samples was 2.2 (range from 1.3 to 3.0), whichshould be representative of emissions from old soil residues.Air sampling surveys in Mexico and Central America, whereDDT is still used, showed a median ratio of 0.56 (range from0.17 to 1.2; Bidleman et al., 2005). The still lower ratios inDurban (median ratio of 0.29, interquartile range of 0.62,n=48), reflect continued DDT use. Despite the low ratios, afraction of DDT and its derivatives is undoubtedly weatheredmaterial.

The ratio of DDD/DDT found at Durban is unusual. Tech-nical grade DDT normally contains only a few percent p,p′-DDD, and weathered residues in well aerated soils usuallycontain more p,p′-DDE than p,p′-DDD (Melnikov, 1971). InDurban, airborne levels of DDD (3-site average of 25±11pgm−3)are almost equal to p,p′-DDT levels, and DDD exceeded DDTlevels in a few samples. This suggests that emissionscannot occur from technical grade DDT alone. Possibly, the‘excess’ p,p′-DDD may be due to the current use of DDT for-mulations that differ from those used decades ago. It should benoted that prior to about 1970, p,p′-DDD was produced as aninsecticide (under the acronym TDE for tetrachloridipheny-lethane, and the trade nameRhothane, among others), despiteits generally lower efficacy compared to p,p′-DDT. Airbornelevels of DDD may reflect volatilization of residues of thisinsecticide remaining in soils. It is also possible that DDT beingused in Africa, which was probably manufactured many yearsago andwhich is unstable in storage, has partially decomposedand formed derivatizaton products.

4.5. Aldrin and dieldrin

Aldrin was found at Nizam (average of 19±17 pg m−3, found inall samples) and Ferndale (8±10 pg m−3, most samples), butonly one detection was found at Wentworth (1.1 pg m−3). Only1–3 detections of dieldrin were made at each site, each with alow concentrations (b0.06 pg m−3) and with slightly over half(54%) in the particulate phase. The trajectories (Fig. 2F) showsource areas that are generally similar to that seen for thechlordanes.

The insecticides aldrin and dieldrin are closely relatedchemically and have similar toxicities. Aldrin was widely usedto control soil insects in corn, potatoes and other crops, and toprotect wooden structures from termites (Fisher, 1999). It wasbanned in all countries decades ago, including South Africa(banned in 1992; Vermeulen et al., 1998). Aldrin is readilyconverted to dieldrin in the environment. However, thepresence of this insecticide at two of the three sites suggestsongoing local uses. Dieldrin was also used to control of soilinsects and termites, as well as wood borers and textile pests(Fisher, 1999) and the Tsetse fly (Sibbald et al., 1986). Muchhigher levels of both aldrin (257±338 pgm−3) and dieldrin (80±80 pg m−3) were measured in 1997–8 at Lake Malawi (Karlssonet al., 2000). Durban levels suggest residual sources in soils, butfollow the global decline.

4.6. Endrin

Wentworth showed trace levels of endrin in three detections,although concentrations were small (b2 pg m−3). Endrin is afoliar insecticide that has been usedmainly on field crops suchas cotton and grains. It was also used as a rodenticide tocontrol mice and voles (Fisher, 1999). Its use has been bannedin South Africa. The levels of endrin seen in this study aresimilar to the low levels at Lake Malawi, 1.1±1.1 pg m−3

(Karlsson et al., 2000).

4.7. Dacthal

Dacthal is a pre-emergence herbicide used for the control ofannual weeds in crops such as onions and broccoli. The fewpublished measurements of Dacthal in air range from a low of3 to 100 pg m−3 in the Great Lakes region and Central Canadawhere its use is very limited, to a high of 4000 to 140,000 pgm−3

in California where this herbicide has been heavily applied(James and Hites, 1999; Rawn andMuir, 1999). Themuch lowerlevels in Durban (average of 0.4±0.5 pg m−3) suggest thatDacthal arises from long-range transport, also suggested asthe explanation for the Canadian monitoring results (Rawnand Muir, 1999).

4.8. Toxaphene

Total toxaphene levels averaged 13.7±5.7 pg m−3 at Nizam,10.7±9.4 pgm−3 atWentworth, and 9.5±2.5 pgm−3 at Ferndale.Most site-to-site differences, at least for the more commonhomologues, were small. This is unsurprising given theexpected regional/global source of these chemicals. Somevariation is expected due to the small samples size. Like theother pesticides, the vapor fraction predominated. The pre-sence of long-range transport from distant sources is indi-cated by higher levels of hexa- and hepta-chloro substitutedforms, while local/fresh use is indicated by higher fractions ofocta and nona-homologues (Hung et al., 2005).

Toxaphene (also known as camphechlor, chlorocamphene,polychlorocamphene and chlorinated camphene) is a non-systemic contact insecticide thatwas introduced in 1949. In theUS, it became one of themost widely used insecticides by 1975until 1982 when most uses were cancelled (ATSDR, 1997).Toxaphene was used primarily on cotton, cereal grains, fruits,nuts, and vegetables, as well as to control ticks and mites inlivestock. In South Africa, toxaphene was withdrawn fromagricultural uses in 1970 and as a stock remedy in 1985.Toxaphene has been banned or uses have been severelyrestricted in over 50 countries. Toxaphene's properties favorlong-range transport in both air and water (Vetter and Oehme,2000). Possibly the closest use is occurring in Egypt or othercotton-growing countries where this pesticide may still beused. We did not locate toxaphene measurements elsewherein Africa.

4.9. Mirex

We did not detect mirex or any products of its derivatization(photo-mirex). Low concentrations (0.51±0.44 pg m−3) werereported at Lake Malawi (Karlsson et al., 2000). Mirex is a

128 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

stomach insecticide that has been used primarily to controlfire ants in the southeastern US, leaf cutters in South America,harvester termites in South Africa, western harvester ants inthe US, andmealybugs inHawaii. It also has been used as a fireretardant in plastics, rubber, paint, paper, and electrical prod-ucts (Fisher, 1999). Mirex was never registered for use in SouthAfrica, but recently has been found in bird eggs collected incentral and southwestern South Africa at uniform but relativelow levels, suggesting contributions from background sources(Bouwman et al., 2007).

4.10. Hexachlorocyclohexanes (HCH)

High concentrations of γ-HCH (lindane) were found. Like otherpesticides, γ-HCH was present largely (90–93%) in the vaporphase, and the highest concentration occurred in late summerat each site (Fig. 4A). Additionally, a strong north-to-southgradient wasmaintained over the year, with concentrations of33±30, 77±54 and 253±276 pg m−3 at Ferndale, Wentworthand Nizam, respectively (Figs. 3, 4). Levels of α-HCH weremuch lower, with comparable levels at Nizam andWentworth(2.2±2.2 and 2.0±1.8 pgm−3), and lower levels at Ferndale (0.5±0.4 pg m−3). The trajectory analysis for HCH was somewhatcomplex, with the higher lindane levels arriving from Ninterior Africa and the N Indian Ocean (31–38% above themean), while the lowest levels arrived from N KZN, S IndianOcean, and the S Atlantic (Fig. 2A). High volume PUF samplingof HCHs is known to have breakthrough problems (Alegriaet al., 2006), thus, reported concentrations represent a lowerbound on the actual vapor phase concentrations.

HCHs have been usedworldwide as an insecticide since the1950s. Technical HCH, containing about 60–70% α-HCH, 5–12%β-HCH, 10–15% γ-HCH, and smaller amounts of five otherisomers, was banned in North America in the 1970s and someyears later in South Africa, although uses still continue inother countries (Garmouma and Poissant, 2004). Use of pureγ-HCH, the isomer with insecticidal properties, continueduntil quite recently, mainly as a seed dressing for corn andcanola, however, most countries phased out γ-HCH by 2004.However, the South African Department of Agriculture (2008)continues to list “Lindaan” and several other products as“gamma-BHC.” Airborne measurements in earlier and mostlynorthern hemisphere studies show levels of γ-HCH thatrange from an average of 9 pg m−3 (South Atlantic) to101 pg m−3 (Anicet, Canada), excepting a very high result atRegina, Canada (1262 pg m−3) associated with very recent andlocal usage (Garmouma and Poissant, 2004). Recent Europeanlevels are 7.6 pg m−3 (EMEP 2002–4 median). The GAPSmeasurement in South Africa for γ-HCH (67 pg m−3) wereabout half those in Durban, but γ-HCH measurement (117 pgm−3) was much higher (Pozo et al., 2006). Previously, the mostextensive measurements of organochlorine pesticides in sub-Sahara Africa were collected at Senga Bay on Lake Malawi inMalawi in 1997–8 (Karlsson et al., 2000). At this location,considered a background site, γ-HCH concentrations aver-aged 25±42 pg m−3 (n=28, 24-h samples). These levels aresimilar to those measured at Ferndale, away from mostindustry and agriculture. Because concentrations of γ-HCHat Nizam far exceeded these levels, local use is suggested,possibly at the several small farms near the airport.

Garmouma and Poissant (2004) report average levels of α-HCH from 9 pg m−3 (south Atlantic) to 145 pg m−3 (Egbert,Canada); the EMEPmedian is 12.5 pgm−3, and the LakeMalawiaverage is 9±9 pg m−3. In Durban, this isomer shows muchlower levels (0.5 to 2.2 pg m−3 across the three sites), probablyreflecting the eliminationof technicalHCHsince the 1989–1996period reported by Garmouma and Possiant (2004) and the1997–8 data in Malawi. In addition, the ratio of α- to γ-HCHacross the three Durban sites (0.017) is much lower thanreported at Lake Malawi and in the Antarctic (Karlsson et al.,2000), also suggesting little use of technical HCH. Becauseα- and γ-HCH concentrations at each Durban site were highlycorrelated (0.53brb0.87), at least some of the α-HCH appearsassociated with the local γ-HCH sources, possibly becausethe insecticide is produced using old technology. However,other sources cannot be ruled out, including volatilization oftechnical HCH remaining in soils, environmental isomeriza-tion of γ-HCH, and long-range transport.

4.11. Pentachloroanisole

Pentachlorophenylmethyl ether (2,3,4,5,6-pentachloroanisole)was detected in all samples. Nizam showed the highest con-centrations (20±13 pg m−3), approximately twice that mea-sured at the other sites. While not a pesticide, this persistentchlorinated aromatic compound is widely distributed at lowlevels in the environment (including foods). We are unsure ofthe source, although pentachloroanisole can be formed duringthe environmental degradation of structurally-related, com-mercially important, ubiquitous and toxic compounds, e.g.,pentachlorophenol and pentachloronitrobenzene (NTP, 1993),and it has been periodically monitored in biota in the US/Canadian Great Lakes. Karlsson et al. (2000) found somewhatlower levels at Lake Malawi almost a decade earlier.

4.12. Source characterization

Local sources of organochlorine compounds in air are suggestedby several factors: high concentrations that exceed global orbackground levels; seasonal patterns of concentrations that arecorrelated to applications; spatial variation among themonitor-ing sites; compositional clues such as low levels of degradationproducts relative to the parent compounds; and trajectoryanalyses that identify and show large differences among pos-sible source areas. The converse of these factors, in general,suggests global sources. As these distinctions are not alwaysclear, each of these factors should be examined. For example,local sources might include fresh uses from agriculture andmalaria-control applications, and possibly residential uses.However, local sources also can include volatilization from soilresidues, and emissions from abandoned and contaminatedstorage sites. Soils at such sites can contain large quantities ofDDT, HCH and other POPs that were manufactured many de-cades ago, even after surface cleanups (Elfvendahl et al., 2004).Temperature-related patternsmay resemble seasonal patterns,with lower temperatures decreasing airborne concentrations asvolatilized compounds partition to particles and water, andincreasing deposition rates, as observed for DDT (Larsson et al.,1995) and HCHs (Garmouma and Poissant, 2004). Local sourcesmay include emissions from soils that are weathered or aged,

129S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

thus resembling the composition of distant/global sources.Finally, local sources that are widely dispersed and common,e.g., waste burning, may result in moderately elevated levelswithout much site-to-site variation. However, source areas forPM10,muchofwhich appears to bewidelydispersedagriculturalburning, did not appear to coincide with the source areas forpesticides (Fig. 2). Additional uncertainty arises from the lack ofinformation regarding the pesticide formulations currently andpreviously used in Africa. Finally, the general lack of data forestablishing global or background levels in the southern hemi-sphere, and Africa in particular, is humbling. Some informationregarding organochlorine sources, including production, use,storage of obsolete pesticides, and ambient levels, is provided ina recent United Nations Environment Program report (UNEP,2002). This report suggests thatwhile overall use of pesticides insub-Saharan Africa is low in comparison to agricultural appli-cations elsewhere, applications can occur at high rates andtherefore contribute towards local impacts and the creation ofpolluted areas. It also identifies themostwidely-usedpersistenttoxic pesticides in the region, i.e., endosulfan, chlordane,lindane, DDT, heptachlor, toxaphene, HCB, and aldrin (in noparticular order), the same chemicals reported here.

While additional monitoring data and information regard-ing current and historical uses of pesticides (specifically thelocations and amounts used) would help to confirm sources,with several caveats we suggest the following: chlordane andlindane arise from both local sources in the DSIB as well asregional/global sources; DDT arises from regional sourceselsewhere in South Africa, Africa and possibly Asia (India andChina); and most of the other long-lived pesticides detected,including γ-nonachlor, hexachlorobenzene and toxaphene,likely represent global sources. The elevated levels of penta-chloroanisole at Nizam may suggest a local source, butuncertainties in measurements must also be recognized.

5. Conclusions

This paper has presentedmonitoring results for a wide range ofpersistent organochlorine compounds using three high volumeair samplers inDurban, South Africa, and it represents themostdetailed study to date of pesticides in air in Africa and some offirst measurements in southern Africa for several compounds,e.g., dacthal and toxaphene.Airbornemonitoringhelps to docu-ment the global distribution of persistent compounds, and theconcentration patterns and trends give information regardingtheir ultimate fate. Because many of the pesticides showedmoderate-to-high inter-site correlation, e.g., the inter-sitecorrelations for Ferndale and Nizam for the pesticides in Fig. 4ranged from 0.74 to 0.86 (n=14 pairs for each comparison), asingle site might be used to establish the temporal variation.Seasonal trends would be much more definitively establishedusing a longer record, e.g., 2 or 3 years of data.

From a public health perspective, themajority of the risk formost people attributable to organochlorine contaminants isgenerally believed to arrive via the ingestion pathway sincethese compounds bioaccumulate in fish, meat and dairyproducts. Unfortunately, there is no information regardinglevels in local foods. In addition to estimating risks, monitoringcan help to identify ongoing uses and emissions, including

applications of pesticides (e.g., lindane) that should have ceasedyears earlier. Finally, pesticide levelsmonitored inmetropolitanDurban, where malaria is not endemic and agriculture isextremely limited, will not reflect levels elsewhere in Kwa-Zulu-Natal province or South Africa where pesticides are orhave been very extensively used.

Acknowledgements

We thank the entire study team that assisted with this work,with special acknowledgements to the following individuals:Siva Chetty at the eThekwini Municipality; Thomas Robins andChunrong Jia, Erika Gywnn, Wei Wang, and Brad Lampe at theDepartment of Environmental Health Sciences, University ofMichigan (USA); and Mike van der Merwe in the Department ofEnvironmental Health Sciences, Durban University of Technol-ogy.Weappreciate theuseof resourcesat theNOAAGreatLakesEnvironmental Research Laboratory in Ann Arbor, MI. Thisprojectwas funded by the eThekwiniMunicipality as part of theMultipoint Plan initiative of national, provincial and localgovernment. Additional financial support was provided by theUSNational Institute of EnvironmentalHealth Sciences, and theUS National Institute of Occupational Safety and Health.

R E F E R E N C E S

Aas W, Breivik K. Co-operative programme for monitoring andevaluation of the long-range transmission of air pollutants inEurope — heavy metals and POP measurements, 2004. Kjeller,Norway: Norwegian Institute for Air Research; 2005.

AlegriaH, BidlemanTF, FigueroaMS.Organochlorinepesticides in theambient air of Chiapas,Mexico. Environ Pollut 2006;140(3): 483–91.

ATSDR. ToxFAQS for toxaphene. Agency for toxic substances anddisease registry, US Department of Health and HumanServices, Public Health Service. Atlanta GA: Agency for ToxicSubstances and Disease Registry; 1997.

ATSDR. Toxicological profile for hexachlorobenzene. Agency fortoxic substances and disease registry, US Department of Healthand Human Services, Public Health Service. Atlanta GA: Agencyfor Toxic Substances and Disease Registry; 2002. September.

Barber JL, Sweetman AJ, vanWijk D, Jones KC. Hexachlorobenzenein the global environment: emissions, levels, distribution,trends and processes. Sci Total Environ 2005;349:1–44.

BawdenK,OrmerodR, StarkeG, ZeiseK.National dioxin programmetechnical report no. 3, inventory of dioxin emissions inAustralia.National dioxins programme & consultancy, Pacific Air andEnvironment Pty Ltd; 2004.

Bidleman T,Wong F, Alegria H. DDT/DDE ratios as indicators of “old”and “new” residues in the atmosphere. SETAC proceedings. 26thannualmeeting inNorthAmerica,Nov.2005, Baltimore,MD;2005.

Bouwman H, Sereda B, Meinhardt HM. Simultaneous presence ofDDT and pyrethroid residues in human breast milk from amalaria endemic area in South Africa. Environ Pollut2006;144:902–17.

Bouwman H, Polder A, Venter B, Skaare JU. Organochlorinecontaminants in cormorant, darter, egret, and ibis eggsfrom South Africa. Chemosphere 2007. doi:10.1016/j.chemosphere.2007.09.057.

Buckland S, Ellis H, Salter R. Organochlorines in New Zealand:ambient concentrations of selected organochlorines in air.Organochlorines programme, ministry for the environmentreport; 1999.

130 S C I E N C E O F T H E T O T A L E N V I R O N M E N T 3 9 7 ( 2 0 0 8 ) 1 1 9 – 1 3 0

Chernyak SM, Rice CP, Quintal RT, Begnoche LJ, Hickey JP, VinyardT. Time trends for organochlorines and polybrominateddiphenyl ethers in rainbow smelt from Lakes Michigan, Huronand Superior, USA. Environ Toxicol Chem 2005;24(7):1632–41.

Draxler RR, Rolph GD. HYSPLIT (Hybrid Single-Particle LagrangianIntegrated Trajectory) model. Silver Spring, MD: NOAA AirResource Laboratory; 2003.

Elfvendahl S, Mihale M, Kishimba MA, Kylin H. Pesticide pollutionremains severe after cleanup of a stockpile of obsoletepesticides at Vikuge, Tanzania. Ambio 2004;33(8):503–8.

eThekwini. eThekwini air quality monitoring network annualreport 2004. Pollution control support section. Durban, SouthAfrica: Ethekwini Health Department; 2004.

Fisher B. Most unwanted persistent organic pollutants. EnvironHealth Perspect 1999;107:A18–25.

Garmouma M, Poissant L. Occurrence, temperature and seasonaltrends of α- and γ-HCH in air (Quebec, Canada). Atmos Environ2004;38:369–82.

Hung H, Blanchard P, Halsall CJ, Bidleman TF, Stern GA, Fellin P,et al. Temporal and spatial variabilities of atmosphericpolychlorinatedbiphenyls (PCBs), organochlorine (OC) pesticidesand polycyclic aromatic hydrocarbons (PAHs) in the CanadianArctic: results from a decade of monitoring. Sci Total Environ2005;342:119–44.

Incorvia-Mattina MJ, Iannucci-Berger W, Dykas L, Pardus J. Impactof long-term weathering, mobility, and land use on chlordaneresidues in soil. Environ Sci Technol 1999;33:2425–31.

Incorvia-Mattina MJ, White J, Eitzer B, Iannucci-Berger W.Cycling of weathered chlordane residues in the environment:compositional and chiral profiles in contiguous soil, vegetationand air compartments. Environ Toxicol Chem 2002;21:281–8.

James RR, Hites RA. Chlorothalonil and dacthal in Great Lakes airand precipitation samples. J Great Lakes Res 1999;25:406–11.

Jaward FM, Barber JL, Booij K, Dachs J, Lohann R, Jones KC.Evidence for dynamic air-water coupling and cycling ofpersistent organic pollutants over the open Atlantic Ocean.Environ Sci Technol 2004;38:2617–25.

Jones K. 2005. Hexachlorobenzene — sources, environmental fateand risk characterization, Science Dossier, Euro Chlor, Brussels,Belgium, Jan. Available: http://www.eurochlor.org/upload/documents/document66.pdf.

Karlsson H, Muir DCG, Teixiera CF, Burniston DA, Trachan WMJ,Hecky RE, et al. Persistent chlorinated pesticides in air, water,and precipitation from the Lake Malawi area, Southern Africa.Environ Sci Tech 2000;34:4490–5.

Keith LH. 1991. Environmental sampling and analysis: a practicalguide. Lewis Publishers, CRC Press Inc., Boca Raton, Florida.

Larsson P, BerglundO, Backe C, Bremle G, Eklov A, Jarnmark C, et al.DDT — fate in tropical and temperate regions.Naturwissenschaften 1995;82:559–61.

Melnikov NN. Chemistry of pesticides. Residue Reviews. Berlin:Springer Verlag; 1971. (Heidelberg, New York, 480 pp.).

NTP. Toxicology and carcinogenesis studies of pentachloroanisole(CAS No. 1825-21-4) in F344 rats and B6C3F1 Mice (FeedStudies). National Toxicology Program. Tech Rep Ser1993;414:1–284 (April).

Offenberg JH, Naumova YY, Turpin B, Eisenreich SJ, Morandi MT,Stock T, et al. Chlordanes in the indoor and outdoor air of threeU.S. cities. Environ Sci Tech 2004;8:2760–8.

Popp P, Brüggemann L, Keil P, Thuβ U, Weiβ H. Chlorobenzenesand hexachlorocyclohexanes (HCHs) in the atmosphere ofBitterfeld and Leipzig (Germany). Chemosphere2000;41:849–55.

Pozo K, Harner T, Wania F, Muir DCG, Jones KC, Barrie LA. Towarda global network for persistent organic pollutants in air:results from the GAPS study. Environ Sci Technol2006;40(16):4867–73.

Preston-Whyte RA, Tyson PD. The atmosphere and weather ofSouthern Africa. Cape Town, South Africa: Oxford UniversityPress; 1989.

Rawn DK, Muir DC. Sources of Chlorpyrifos and Dacthal to a smallCanadian prairie watershed. Environ Sci Technol1999;33:3317–23.

Reid K, D'Sa D. The double edged sword: advocacy and lobbying inthe environmental sector. Crit. Dialogue - Public Particip Rev2005;2(1) http://www.h-net.org/~esati/sdcea/pressreleases.html.

Shen L Wania F, Lei YD, Teixeira C, Muir DCG, Bidleman TF.Atmospheric distribution and long-range transport behavior oforganochlorine pesticides in North America. Environ Sci Tech2005;39(2):409–20.

Sibbald RR, Connell AD, Butler AC, Naidoo P, Dunn JD. A limitedcollaborative investigation of the occurrence of dieldrin inselected biota in the Durban area. S Afr J Sci 1986;82:319–20.

Simpson AJ, McGee OS. 1994. Analysis of the fumigation effect onpollutants over Pietermaritzburg. [Online] Accessed on Feb 24,2001, Available at: http://www.egs.uct.ac.za/sagj/simpson.htm.

South African Department of Agriculture, 2008, Banned andrestricted substances in the Republic of South Africa, http://www.nda.agric.za/act36/Banned%20and%20restricted.htm,accessed 1/18/08.

Strandberg B, Bavel BV, Bergqvist PA, Broman D, Ishaq R, Naf C,et al. Occurrence, sedimentation, and spatial variations oforganochlorine contaminants in settling particulate matterand sediments in the northern part of the Baltic Sea. EnvironSci Technol 1998;32:1754–9.

Tyson PD, Preston-Whyte RA. Observations ofregional topographically induced wind systems inNatal. J Appl Met 1972;11:643–50.

UNEP. Regionally based assessment of persistent toxic substanceschemicals. Sub-Saharan regional report. Châtelaine, Switzerland:United Nations Environment Program; 2002. Dec.

US EPA. US EPA RED FACTS, Dicofol, Office of Pesticide Programs,US EPA, Washington, 1998; 1998 (7 pp.).

Vermeulen JB, Grobber HH, vanZyb K. A guide to use of herbicides.16th ed. National Department of Agriculture, Johannesburg,South Africa; 1998.

Vetter W, Oehme M. In: Paasivirta J, editor. The handbook ofenvironmental chemistry. Part K, New Types of PersistentHalogenated CompoundsHeidelberg, Germany: Springer; 2000.p. 237–87.

Wania F, Mackay D. Global fractionation and cold condensation oflow volatility organochlorine compounds in polar regions.Ambio 1993;22:10–8.

Wekker SFJ, Zhong S, Fast JD, Whiteman CD. A numerical study ofthe thermally driven plain-to-basin wind over idealized basintopographies. J Appl Met 1998;37(6):606–22.