Environ Monit Assess DOI 10.1007/s10661-014-3809-3

Assessment of bioavailable fraction of POPS in surface water bodies in Johannesburg City, South Africa, using passive samplers: an initial assessment Robert Amdany & Luke Chimuka & Ewa Cukrowska & Petr Kukučka & Jiří Kohoutek & Peter Tölgyessy & Branislav Vrana Received: 2 November 2013 / Accepted: 6 May 2014 # Springer International Publishing Switzerland 2014

Abstract In this study, the semipermeable membrane device (SPMD) passive samplers were used to determine freely dissolved concentrations of polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs) and organochlorine pesticides (OCPs) in selected water bodies situated in and around Johannesburg City, South Africa. The devices were deployed for 14 days at each sampling site in spring and summer of 2011. Time weighted average (TWA) concentrations of the waterborne contaminants were calculated from the amounts of analytes accumulated in the passive samplers. In the area of interest, concentrations of analytes in water ranged from 33.5 to 126.8 ng l−1 for PAHs, from 20.9 to 120.9 pg l−1 for PCBs and from 0.2 to 36.9 ng l−1 for OCPs. Chlorinated pesticides were mainly composed of hexachlorocyclohexanes (HCHs) (0.15–36.9 ng l−1) and dichlorodiphenyltrichloromethane (DDT) with its metabolites (0.03–0.55 ng l−1). By applying diagnostic ratios of certain PAHs, identification of possible sources of the contaminants in the various sampling sites was R. Amdany : L. Chimuka (*) : E. Cukrowska Molecular Sciences Institute, School of Chemistry, University of the Witwatersrand, P/Bag 3, WITS, Johannesburg 2050, South Africa e-mail: [email protected] P. Kukučka : J. Kohoutek : B. Vrana Research Centre for Toxic Compounds in the Environment (RECETOX), Masaryk University, Kamenice 573/5, CZ-625 00 Brno, Czech Republic P. Tölgyessy Water Research Institute, Nabr. Arm. Gen. L. Svobodu 5, 812 49 Bratislava, Slovakia

performed. These ratios were generally inclined towards pyrogenic sources of pollution by PAHs in all study sites except in the Centurion River (CR), Centurion Lake (CL) and Airport River (AUP) that indicated petrogenic origins. This study highlights further need to map up the temporal and spatial variations of these POPs using passive samplers. Keywords Free dissolved concentration . Passive sampling devices . Hydrophobic organic compounds . Monitoring . Passive sampling . SPMDs

Introduction Water systems that have roots in urbanised areas are normally prone to severe contamination by an array of pollutants that include hydrophobic organic contaminants (HOCs). Pollution may be caused by current and/or previous industrial activities or both. Such water resources need to be secured for the benefit of current and future generations. Assessment of the pollution levels and distribution of the contaminants in water systems can be achieved by employing sound monitoring practices using a variety of available tools and techniques. Grab sampling has traditionally been applied in the determination of HOCs in water. However, successful monitoring is hampered by their existence at very low concentrations in water phase, in addition to frequent temporal changes. Increased sampling frequency, use of large sample volumes, installing automatic samplers and applying more sensitive analytical

Environ Monit Assess

techniques are possible solutions, but they come with cost implications. Passive sampling devices (PSDs) have shown promise as better alternatives since they permit unattended large volume- and time-integrated sampling, which compensate for fluctuating concentrations and also give lower detection limits (Harman et al. 2008; Vrana et al. 2014). Use of passive samplers is also advantageous because only the freely dissolved concentration of the analyte in water is sampled. This fraction of the contaminant is critical for the assessment of its bioavailability and fate in the aquatic environment and the risk associated with exposure of aquatic organisms to these contaminants. Among PSDs, the semipermeable membrane devices (SPMDs) have been successfully used as quantitative tools to assess concentrations of HOCs in the waters of various aquatic ecosystems (Huckins et al. 1993; Lu et al. 2002; Vrana et al. 2005, 2014). SPMDs passively accumulate lipophilic organic contaminants by mimicking biological membranes in its ability to allow selective diffusion of the compounds. Typically, organics with partition coefficients (log Kow) higher than 3 are suitable for extraction by this technique (Huckins et al. 1993; Vrana et al. 2005). In the current study, SPMDs were employed for the initial assessment of the bioavailable fractions of polycyclic aromatic hydrocarbons (PAHs), polychlorinated biphenyls (PCBs) and organochlorine pesticides (OCPs) in the water columns of streams and rivers originating from Johannesburg City, South Africa. Further investigation of the pollution levels in water bodies that receive water from the urban streams and rivers was also undertaken. So far, very little studies have been reported that have investigated the presence of these POPs in water bodies in greater Johannesburg area, South Africa (Sibali et al. 2008; Sibiya et al. 2012, 2013a;). The reported studies have looked at the total concentrations and only for a few PAHs (Sibiya et al. 2012, 2013a) and organochlorine pesticides (Sibali et al. 2008).

Materials and methods Chemicals and reagents The 16 US EPA PAH standards with purities>97 % pure were purchased from Sigma-Aldrich Chemie GmbH, (Steinheim, Germany). Performance reference compounds (PRCs): d10-acenaphthene, d10-fluorene, d10-phenanthrene and d10-pyrene, as well as recovery standards d8-

naphthalene, d10-anthracene, d12-fluoranthene, d12benzo(a)anthracene, d12-benzo(k)fluoranthene, d12benzo(g,h,i)pyrene, PCB 30, PCB 185 and d6-gamma hexachlorocyclohexane (HCH) were purchased from Dr. Ehrenstorfer GmbH (Augsburg, Germany). Internal standards (PCB 121 and terphenyl) for instrumental analysis were also purchased from Dr. Ehrenstorfer GmbH. Sulphuric acid (98 %) and hydrochloric acid (36 %) were purchased from Merck (Darmstadt, Germany). Triolein (97 %,) was purchased from Sigma Aldrich, (Ghent, Belgium), while silica gel 60 was from Merck (Darmstadt, Germany). High-purity (>99 %) n-hexane, dichloromethane and trichloromethane were bought from Sigma-Aldrich (Prague, Czech Republic). Reagent water was drawn from a Milli-Q water system (Millipore, Bedford, MA, USA). SPMD sampler preparation and deployment Standard size SPMDs in the dimensions 2.54×91.4 cm, 460 cm2 external surface area were prepared from LDPE membranes (Brentwood plastics, MO, USA) and filled with 1 ml of high-purity triolein (97 % pure) to give a final total sampler volume of 4.95 ml. Initially, 100-cm long portions of the tubes were cut off from the roll before inserting them into pre-cleaned, dry glass bottles using a pair of blunt tweezers. Cleaning was done twice by soaking them overnight in hexane with the aim of removing organic contaminants. After airdrying, the membranes were heat-sealed at one end to form a loop. Each SPMD was spiked (a solution in nhexane was spiked to SPMD using a GC syringe) with individual PRCs at a concentration of 2 μg sampler−1 (Vrana et al. 2014). The LDPE membrane was closed using a thermal seal (Impulse sealer ME-300 HI, Mercier Corporation). The devices were stored in airtight sealed metal cans at −20 °C awaiting deployment. Thereafter, the stainless steel housings containing the SPMDs were lowered about 40 cm below the surface of a river or dam bank. Its end was then tied to nearby branch of tree using a nylon string. At Hartbeespoort Dam (HD), the passive samplers were tied to the bottom of the floating bridges used for recreation purposes in the same way using a string. Samplers were deployed for 14 days; the more commonly used deployment time (Vrana et al. 2014). Sites for sampler deployment were based on our previous study in the same area (Sibiya et al. 2013b) and are described in Table 1. The sampling points are also shown in Fig. 1.

Environ Monit Assess Table 1 Description of sampling sites Sampling site

Symbol

Water body

Longitude

Latitude

Ifafi (Hartbeespoort Dam)

IFA

Dam

25° 45′ 09.97″ S

27° 53′ 04.39″ E

Juskei River 1

JR 1

River

26° 01′ 07.49″ S

28° 05′ 34.69″ E

Juskei River 2

JR 2

River

26° 00′ 25.30″ S

28° 04′ 45.74″ E

Eagles (Hartbeespoort Dam)

EGL

Dam

25° 44′ 56.29″ S

27° 50′ 06.18″ E

Homestead Lake

HSL

Dam

26° 10′ 25.64″ S

28° 17′ 04.71″ E

Airport River

AUP

River

26° 08′ 29.74″ S

28° 17′ 04.71″ E

Centurion Lake

CL

Dam

25° 51′ 55.41″ S

28° 12′ 23.36″ E

Centurion River

CR

River

25° 51′ 40.01″ S

28° 11′ 22.93″ E

Sampling sites Jukskei Rivers The Jukskei River (JR) is one of the river catchments in the Johannesburg metropolis that covers over 800 km2 (Campbell 1996). The source of the river can be traced to the Bruma Lake situated at the foot of the Witwatersrand area. The river eventually empties its water into the Hartbeespoort Dam after merging with the Crocodile River downstream. The Jukskei River (JR) meanders northwards through a number of residential areas such

Fig. 1 Map of the sampling sites

as the densely populated Alexandra Township. Two sampling sites were chosen along the Jukskei River (JR 1 and JR 2) for the deployment of SPMDs. Centurion Lake and Centurion River The Centurion Lake (CL) and Centurion River (CR) form part of the Hennops, a relatively small perennial river, that originate from a marshy area situated a few kilometres east of Kempton Park, Johannesburg (Torien and Walmsley 1979). Further downstream, the river receives treated effluent from a number of wastewater

Environ Monit Assess

treatment plants. Kempton Park area is home to several industries that release their effluent into the river via the treatment plants. The Centurion Lake is surrounded by vibrant business installations such as shopping malls, hotels, recreational facilities and garages. Hartbeespoort Dam Two sites were located in the Hartbeespoort Dam: Ifafi (IFA) and Eagles (EGL). The dam is located about 37 km west of Pretoria and along the Crocodile River in North West Province of South Africa. The water body is a 20 km 2 water reservoir sandwiched between the Megaliesberg mountain range in the Highveld region of northern South Africa (Hely-Hutchinson and Schumann 1997; Nyoni et al. 2011). The reservoir receives water from an approximately 4,100 km2 area all the way from Johannesburg City via the Jukskei and Hennops Rivers that flow into the Crocodile River. This accounts for about 90 % of the dam’s water inlet supply with rain water being the major source (Harding et al. 2004). The five catchment basins of the dam are from west to east, the Megalies/Skeerpoort, the Crocodile, the Jukskei, the Hennops and the Swartspruit basin (Van Rei 1987). Homestead Lake and Airport River Both Homestead Lake (HSL) and Airport River (AUP) are located about 28 and 35 km, respectively, east of Johannesburg City central business district and a few kilometres from the O.R. Tambo international airport. The origin of the AUP is a swampy area situated at the periphery of the airport. It flows downstream in an easterly direction, passing through a few residential areas before discharging its water into a man-made reservoir called HSL. This dam is surrounded by many residential developments. The HSL also receives water from a small stream originating from the southern part of the airport and about 5 km west of the dam. After exiting the dam, the river flows towards the east rand area of Johannesburg. Extraction of SPMDs Debris, particulates and biofouling were removed from the surfaces of retrieved SPMDs using a stream of tap water before briefly immersing in diluted (10 %) hydrochloric acid to rid them of adsorbed carbonates. Further washing of the

samplers with deionised water and air-drying at room temperature followed before placement of each device in a 250-ml glass container with a ground joint glass stopper. One hundred milliters of n-hexane was added into each container to fully immerse the SPMD and spiking with surrogate standards in hexane (0.5 μg sampler−1 of each compound) done. Dialytic extraction of analytes was carried out over a 24-h period at room temperature and in the dark. After this period, dialysates were transferred into clean, labelled glass containers and fresh batches of 100 ml of n-hexane added to the samplers and the process repeated. The extracts were combined and reduced to about 10 ml at 40 °C using a rotary evaporator (Heidolph Laborata 4000, Germany) and further concentrated to the last drop with a gentle stream of nitrogen gas and, thereafter, reconstituted in 1 ml of trichloromethane. Further processing by gel permeation chromatography (GPC) was done to remove triolein and sulphur contaminants prior to instrumental analysis. One thousand microliters of the extract was introduced into a GPC system equipped with a high-pressure pump (HPP5001) and a fraction collector (ECOM, Prague, Czech Republic) and fractionation achieved using a Gel 5 μm 50 Å, 7.5 × 300 mm, column (Agilent PL). Dichloromethane acted as the mobile phase flowing at 0.6 ml min−1. Target analytes were collected in the fractions that eluted as from 18.3 to 41.7 min. Prior to eluent volume reduction to near dryness using nitrogen gas, a solvent keeper (0.1 ml of n-nonane) was added. The sample was finally reconstituted in 1 ml of n-hexane and subjected to activated silica gel cleanup. For PAHs, each column was packed with 5 g of activated silica gel (prepared by drying at 120 °C for 8 h). Conditioning of the column was done by flushing it with 10 ml of n-hexane and the analytes of interest eluted using 20 ml of dichloromethane after sample introduction. The eluate was evaporated to 10 ml at 40 °C and reduced further to 1 ml with nitrogen gas. Sulphuric acid-modified activated silica gel (mixture of 50 g freshly prepared activated silica gel and 33 ml of concentrated sulphuric acid, 98 %) was used to clean PCBs and OCPs. Subsequent elution of the analytes was done with 30 ml dichloromethane. After evaporation and further concentration to 1 ml, terphenyl or PCB 121 internal standards were added to the samples. GCMS analysis of the compounds followed.

Environ Monit Assess

Instrumentation

Water-dissolved concentrations of compounds

Prior to analysis of all samples, calibration using standards with concentrations ranging from 0.01 to 0.200 μg ml−1 was done. For PAHs, separation was achieved using a HP-5MS capillary column with the dimensions 30 m × 0.25 mm internal diameter, 0.25 μm film thickness and helium as the carrier gas flowing at 1.9 ml min−1. Working conditions: splitless injection of 1 μl sample at 250 °C. Column temperature programme 70 °C (0.5 min hold) then ramped at 25 °C min−1 to 150 °C followed by 3 °C min−1 to 200 °C and finally increased at 8 °C min−1 to 280 °C and held for 20 min. Detection of separated analytes was done using a 5971 MS system (Agilent Technologies, USA) set at 320 °C and 70 eV electron impact ionisation. Selected ion monitoring mode was used in the measurements and two to three characteristic ions were chosen for detection and quantification of each compound. The ion source, the transfer line and the quadrupole temperatures were maintained at 230, 280 and 150 °C, respectively. Using external calibration methods, analyte concentrations in the samples were calculated based on the peak areas of the highest characteristic ion in the mass spectrum of the compounds. Recoveries of surrogate standards introduced into the sampler containers prior to dialytic extraction were used to correct such concentrations. Analysis of PCBs and OCPs in the samples was done on a 6,890 N GC (Agilent technologies, Santa Clara, USA) linked to a Quattro MicroGC MS (Waters, Micromass, UK) operated in EI+ mode was used. Chromatographic separation of target analytes was achieved using a 60 m×0.25 mm× 0.25 um DB5-MS column (Agilent J&W, USA) with the column flow rate of carrier gas (helium) maintained at 1.5 ml min−1. The inlet was operated in the splitless mode at 280 °C and 1 μl sample loaded. For each compound analysed, a minimum of two MRM transitions were recorded. The column temperature was initially 80 °C and held for 1 min, then ramped at 15 °C min−1 to 180 °C and finally 5 °C min−1 to 300 °C (5 min). TargetLynx software (Waters, Micromass, UK) was applied in processing the raw data. Contaminations that may have occurred during sampler fabrication, deployment and/or retrieval were corrected using fabrication and field blanks.

Since amounts of analytes absorbed by the samplers follow a first-order approach to equilibrium, waterdissolved concentrations were determined from the quantities sequestered (Ns) by SPMDs, compoundspecific in situ sampling rates (RS) and their samplerwater partition coefficients (Ksw): Cw ¼

Ns K sw V s ½1−expð−Rs t=ðK sw V s Þފ

ð1Þ

where VS =SPMD volume and t=exposure time. The dissipation of PRCs also obeys first-order kinetics. The nonlinear least squares method (Booij and Smedes 2010) was adopted in the estimation of Rs based on the fraction (f) of individual PRCs that remained in the SPMD after exposure as a continuous function of their Ksw, with RS as an adjustable parameter.   Rs t f ¼ exp ð2Þ K SW V S where, f = NPRC/N 0,PRC and N 0,PRC = PRCs at t = 0, NPRC =PRCs at t=14 days. With the assumption that uptake is controlled by the aqueous boundary layer, Eq. (3) (Rusina et al. 2010) was substituted in Eq. (2) enabling the estimation of sampling rates of individual compounds in the higher hydrophobicity range. RS ¼ FAM −0:47

ð3Þ

where M=molar mass of compound, A=SPMD surface area (460 cm2) and F=regression coefficient that was optimised using nonlinear least squares method for estimating sampling rates. Ksw values were intrapolated from the empirical Eq. (4) (Huckins et al. 2006). Log Kow values were obtained from various literatures (Vrana et al. 2014). Log K sw ¼ −01618ðlogK ow Þ2 þ 2:321logK ow −2:61

ð4Þ

Results and discussion Occurrence of PAHs, PCBs and OCPs in the SPMDs Amounts of PAHs accumulated in SPMDs after the 14day exposure period in the various sampling sites are shown in Table 2, while for PCBs and OCPs are presented in Table 3. Field blank SPMDs were devoid of

Environ Monit Assess Table 2 Mean concentrations of PAHs in SPMDs (ng sampler−1) sequestered from different sample sites (n=3) Compound

Sampling site IFA

JR 1

JR 2

EGL

HSL

AUP

CL

CR

Naphthalene

268±41

247±56

262±16

283±61

389±88

307±26

338±71

148±22

Acenaphthylene

156±30

156±18

158±29

150±12

164±26

197±28

132±11

55±4

Acenaphthene

34±5

38±7

124±6

32±7

38±9

334±6

47±4

74±8

Fluorene

84±15

77±4

355±8

65±11

82±20

71±11

36±8

213±22

Phenanthrene

144±12

246±5

1,192±97

132±15

138±18

174±18

1,220±98

Anthracene

308±14

287±83

198±11

102±14

49±10

179±22

121±9

31±5

Fluoranthene

127±10

424±47

999±35

156±37

61±15

83±21

118±3

260±19

Pyrene

96±13

367±38

825±91

115±12

42±5

97±19

279±35

260±22

Benz[a]anthracene

32±8

62±7

119±6

32±2

29±0

31±2

189±14

18±4

Chrysene

44±9

133±18

203±28

46±8

34±4

42±5

11±2

31±4

Benzo[b]fluoranthene

38±3

57±4

86±4

34±1

39±6

36±2

17±4

9±0

Benzo[k]fluoranthene

25±5

62±3

86±17

38±4

37±2

71±5

5±1

5±1

Benzo[a]pyrene

30±4

37±13

55±4

79±12

48±6

Indeno[1,2,3-cd]pyrene

38±3

41±2

48±10

38±5

37±3

Dibenz[a,h]anthracene Benzo[ghi]perylene ΣPAHs

ND 40±4 1,464±60

ND

ND

45±3

56±5

2,279±121

4,766±147

ND 36±2 1,335±79

ND 36±5 1,223±98

ND 38±0 ND 37±2 1,394±57

425±13

11±1

4±1

4±1

8±0

ND 18±3 2,543±128

ND ND 1,541±46

ND not detected

quantifiable amounts of PAHs, PCBs and OCPs. Prior to dialytic extraction of analytes in the samplers, recovery standards were introduced. Information obtained from their recoveries was then used to adjust the concentrations of the target compounds. Compounds of interest showed good recoveries that ranged from 55 to 115 % for PAHs, 76 to 103 % for PCBs and 69 to 111 % for OCPs. Comparable quantities of the analytes were obtained from triplicate SPMDs deployed in the same site. Relative percent differences between such replicates in all sites were not greater than 25 % for PAHs, except for anthracene at sampling site JR 1 (29 %) and benzo[k]fluoranthene at sample site AUP (34 %). For PCBs, these differences did not exceed 25 %. Save for α-HCH recorded in sampling site JR 2 (34 %), all OCPs exhibited a relative percent difference of less than 21 % between replicates. In this study, summed-up amounts of SPMDsequestered analytes ranged from 1,223 to 4,766 ng sampler−1 for PAHs, 6.5 to 76.1 ng sampler−1 for PCBs and 4.5 to 921.6 ng sampler−1 for OCPs. The chlorinated pesticides were primarily composed of HCHs (3.0 to 870.0 ng sampler−1) and dichlorodiphenyltrichloromethane (DDT) and its metabolites (1.5 to 86.5 ng sampler−1).

The use of PRCs (D-PAHs) for other compounds other than PAHs is justifiable. Since hydrophobic compounds with log Kow >4 are accumulated in SPMD under water boundary layer control (WBL), sampling rate is determined by diffusion in water. Diffusion coefficients in water of PAHs, PCBs and OCPs are a weak function of molecular weight (Eq. 3) (Rusina et al. 2010). Since diffusion in water is assumed not to be affected by parameters other than molecular volume/weight, sampling rates of all compounds can be estimated using Eq. 3. Water-dissolved concentrations of the analytes Rs values for individual compounds were determined using Eq. 3. Table 4 presents the PRC-derived Rs values for fluorene resulting from SPMD field deployment at the various sampling sites. These values ranged from a low of 1.0 l d−1 at HSL in October 2011 to a high of 26.1 l d−1 at JR 2 in December 2011. Differences in Rs values at different sites may be attributed to variations in water flow velocities, in agreement with assumption of water boundary layer uptake. Although no flow velocities were measured, dams generally had much lower

Environ Monit Assess Table 3 Mean concentrations of PCBs and OCPs in SPMDs (ng sampler−1) recorded in the different sample sites (n=3) Sampling site Compound

IFA

JR 1

JR 2

EGL

HSL

AUP

PCBs PCB 28

34.3±5.5

44.1±7.8

68.1±12

32.0±6.3

PCB 52

1.2±0.2

2.3±0.5

2.4±0.4

0.8±0.1

5.9±1 0.3±0.0

3.2±0.2 2.4±0.1

PCB101

0.6±0.1

1.1±0.3

1.1±0.1

0.3±0.0

0.1±0.0

2.0±0.0

PCB 118

0.2±0.0

0.5±0.1

0.8±0.3

0.2±0.0

ND

ND

PCB 138

0.6±0.0

0.9±0.2

1.3±0.2

0.5±0.1

0.2

3.1±0.1

PCB 153

0.6±0.1

1.2±0.1

1.4±0.2

0.4±0.1

ND

5.1±0.1

PCB 180

0.7±0.1

0.5±0.1

1.0±0.1

0.4±0.0

ND

38.2±0.5

50.6±0.9

76.1±0.6

34.6±0.6

ΣPCBs

6.5±0.1

4.3±0.0 20.1±0.2

OCPs HCHs α-HCH

44.4±8

120.4±41

131.5±26

43.2±7

0.8±0.2

1.2±0.3

β-HCH

127.9±16

626.8±98

630.2±82

114.1±20

0.8±0.1

0.7±0.1

γ-HCH

7.0±1.1

16.5±3.7

13.6±2.2

5.9±1.2

1.4±0.2

1.2±0.2

δ-HCH

2.6±0.4

46.5±4.0

65.9±13.4

2.3±0.0

ND

ND

ε-HCH

6.7±1.3

11.4±1.9

28.8±5.6

7.5±1.3

ND

ND

188.5±10

821.6±107

870.0±87

173.0±21

3.0±0.3

3.1±0.4

0.7±0.1

0.8±0.2

1.5±0.2

0.6±0.1

0.1±0.0

0.1±0.0

p,p′-DDE

7.6±1.3

8.0±1.6

14.4±2.7

5.5±1.0

0.5±0.1

2.1±0.4

o,p′-DDD

20.3±4.8

3.2±0.5

5.2±0.5

28.6±5.5

0.1±0.0

0.6±0.1

p,p′-DDD

55.8±3.4

17.7±3.0

26.6±1.4

47.2±8.1

0.5±0.0

1.5±0.2

o,p′-DDT

1.5±0.0

3.0±0.7

1.8±0.4

1.8±0.3

0.3±0.0

0.5±0.1

p,p′-DDT

0.6±0.1

11.1±2.6

2.1±0.2

0.6±0.0

ND

0.6±0.1

ΣHCHs DDX o,p′-DDE

ΣDDX

86.5±6.0

43.8±4.4

51.6±3.1

84.3±9.8

1.5±0.1

5.4±0.5

ΣOCPs

275±12

865.4±107

921.6±88

257.3±23

4.5±0.3

8.5±0.6

ND not detected All summed numbers are in italic

velocities compared to those samplers deployed on river banks. Other contributors to differences in sampling rates include temperature, biofilm infestation and deposition of particulates on the surface of sampler (Baxter 1990; Cailleaud et al. 2007; Brandli et al. 2008; Booij and Smedes 2010). South Africa has got four seasons with summer starting from mid-October to midFebruary and is very hot characterised by afternoon thunderstorms. Autumn is from mid-February to midApril with little rain and not very hot. Winter starts from May to July, while spring is from August to midOctober. Most of the sampling was done in spring (Table 4) where there is little or no rainfall and is beginning to get hot.

PAHs Estimation of free dissolved concentrations of the PAHs in water based on the amounts accumulated in deployed SPMDs are presented in Table 5. Since the sampling was not done at the same time and season, it is not easy to compare the results for spatial trends. Total analyte concentrations by site varied from 22.1 ng l−1 at RC to 126.7 ng l−1 at HSL. The high concentration of PAHs at this site could be linked to previous reported oil spill in the upstream of the Airport River. Airport River (AUP) flows into the Homestead Lake, and this suggests that it is acting as a recipient of PAHs. In the same way, the concentrations of PAHs in the Centurion Lake were

Environ Monit Assess Table 4 Description of the sampling campaign at the sites Sampling site

Season

Exposure period Start

IFA

Exposure (days)

Water temperature (°C)

SPMD-Sampling rate RS (l d−1)

End

Spring

19

19.2

JR 1

2 September 16 September 14 2011 2011 Summer 3 December 2011 17 December 2011 14

21

19.4

JR 2

Summer 3 December 2011 17 December 2011 14

21

26.1

EGL

Spring

18.3

HSL AUP

16 September 2011 20 October 2011

14

19

Spring

2 September 2011 6 October 2011

14

20

1.0

Spring

06 October 2011 20 October 2011

14

20

20.0

CL

Spring

12 August 2011

26 August 2011

14

18

10.7

CR

Spring

12 August 2011

26 August 2011

14

18

18.8

much higher than those in the Centurion River. This again may suggest that the dam is acting as a recipient and perhaps a sink of PAHs. A study of PAHs in sediments in the same area found high concentration levels (Sibiya et al. 2013b). The concentration of PAHs in sediments at the Centurion Lake ranged from 61 to 1,690 μg kg−1 and 84 to 1,545 μg kg−1 at the Homestead Lake (Sibiya et al. 2013b). The reported concentrations in Table 5 for PAHs are comparable to those reported by Karacık et al. (2013) (8.36–76.68 ng l−1) and Wang and coworkers (2009) (19.14–97.17 ng l−1). They were also comparable to those obtained by Vrana et al. (2014) in the Danube River (13–72 ng L−1). However, they were significantly higher than what Allan and Ranneklev (2011) (0.033–9.3 ng l−1) obtained in the Alna River, Norway. Evidently, water phase PAH concentrations of individual compounds (Fig. 2) generally reflected the trend exhibited by the cumulative concentrations at any given sampling site. The smaller molecular weight compounds (≤ four rings) accounted for the highest percentage (77.6 % at HSL to 96.5 % at AUP) of total PAHs in the water phase. Their relatively higher water solubilities as indicated by lower log Kow values enhance their availability and, hence, uptake by the samplers. On the other hand, strong hydrophobicity of larger molecular weight PAHs encourages increased sorption to larger particulates and colloids in the water column resulting in diminished availability. PCBs Freely dissolved PCB levels in the waters of the various sampling sites are presented in Table 5. Estimated water

phase concentrations were in the low picograms per liter range. Sum of seven indicator PCB congeners varied between 21 pg l−1 at AUP and 121 pg l−1 at HSL. These values were about three orders of magnitude lower than for PAHs and up to two orders lower than OCPs. Of the many PCB congeners known, seven of them (PCB 28, PCB 52, PCB 101, PCB 118, PCB 138, PCB 153 and PCB 180) were quantifiable in most of the sites, with the lesschlorinated PCBs predominating (up to 89 %). Figure 3 presents concentrations of some of the PCB congeners. Residue levels obtained were lower than those reported by Allan and Ranneklev (2011), Liu et al. (2013) and Cailleaud et al. (2007) but generally comparable to those obtained by Wang et al. (2009) (66–519 pg l−1). PCBs enter the environment mainly through volatilisation from in-use and disposed equipment or as re-emissions from soils (Wang et al. 2007). Although these compounds were never produced in South Africa, PCB oils as well as equipment containing such oils were imported for use mainly for electricity generation and in manufacturing industries. However, as in many other countries, PCBs are currently outlawed in the country, and their presence in the environment is attributed to previous applications, since these compounds are persistent organic pollutants. Old electricity transformers contained PCBs, but these are now being phased out by Eskom, a South African electricity generation and supply company. Although sampling sites IFA and EGL are located in an area devoid of major industrial activities, it still recorded significant quantities (33.6 and 27.1 pg l−1, respectively) of the contaminants. Apparently, most of

Environ Monit Assess Table 5 Estimated dissolved water concentrations (ng l−1) of PAHs, PCBs and OCPs at the various sampling sites Compound

Sampling site IFA

JR 1

JR 2

EGL

HSL

AUP

CL

CR

PAHs Naphthalene

37.205

34.705

36.813

61.185

43.135

47.491

Acenaphthylene

3.832

3.795

3.828

38.92 4.084

13.307

4.777

3.279

20.795 1.338

Acenaphthene

1.037

1.154

3.764

0.941

3.194

1.017

1.439

2.247

Fluorene

1.349

1.157

5.280

0.905

6.422

1.063

0.595

3.200

Phenanthrene

1.198

1.699

7.487

0.900

10.465

1.171

11.773

2.942

Anthracene

2.838

1.656

1.306

0.581

3.743

0.918

1.200

0.224

Fluoranthene

0.755

1.938

3.757

0.483

4.753

0.363

0.909

1.194

Pyrene

0.579

1.694

3.152

0.429

3.265

0.427

2.164

1.207

Benz[a]anthracene

0.185

0.276

0.423

0.130

2.409

0.131

1.470

0.081

Chrysene

0.256

0.592

0.726

0.179

2.822

0.178

0.086

0.139

Benzo[b]fluoranthene

0.268

0.263

0.317

0.151

3.370

0.157

0.139

0.042

Benzo[k]fluoranthene

0.141

0.284

0.313

0.198

3.182

0.319

0.041

0.023

Benzo[a]pyrene

0.195

0.171

0.200

0.189

2.023

0.089

0.019

Indeno[1,2,3-cd]pyrene Dibenz[a,h]anthracene

0.236 ND

Benzo[ghi]perylene ΣPAHs

0.247 50.32

0.196 ND 0.214 49.79

0.182 ND 0.212 67.76

0.172 ND 0.165 48.43

3.361 ND 3.27 126.78

ND 0.174 ND 0.167 53.99

0.034 ND 0.152 70.86

0.039 ND ND 33.49

OCPs HCHs α-HCH

1.884

5.106

5.576

1.832

0.092

0.051

β-HCH

5.737

28.103

28.254

5.116

0.094

0.031

γ-HCH

0.371

0.874

0.721

0.313

δ-HCH

0.046

0.790

1.095

0.039

ND

ND

ND

ND

0.170

0.064

ε-HCH

0.301

0.511

1.291

0.336

ΣHCHs

8.339

35.385

36.937

7.635

0.356

o,p′-DDE

0.004

0.004

0.006

0.003

0.010

p,p′-DDE

0.048

0.040

0.055

0.026

0.048

0.010

o,p′-DDD

0.129

0.016

0.020

0.133

0.010

0.003

p,p′-DDD

0.354

0.090

0.102

0.227

0.048

0.007

o,p′-DDT

0.010

0.016

0.007

0.009

0.030

0.003

p,p′-DDT

0.004

0.059

0.008

0.003

ΣDDTs

0.549

0.225

0.198

0.401

0.146

0.026

ΣOCPs

8.888

35.610

37.135

8.036

0.502

0.172

0.034

0.058

0.074

0.027

0.121

0.021

0.146

DDX

ND

ND

0.003

PCBs ΣPCBs ND not detected All summed numbers are in italic

the water at the sites is supplied through the Crocodile River (Harding et al. 2004) whose major tributaries include the Jukskei and Hennops Rivers. The origins

of the two rivers can be traced to the outskirts of Johannesburg City—a probable source. This is especially reinforced by the closeness in the estimated

Environ Monit Assess

Fig. 2 Estimated water-dissolved PAH concentrations of some individual PAHs in the sampling sites

contaminant concentrations at IFA (34 pg l−1) and EGL (27 pg l−1) sampling sites (both in the same water body). The lighter-molecular-weight PCB congeners are usually more prone to atmospheric transport (Ockenden et al. 2003) and volatilisation (Wang et al. 2007). However, presence of larger-molecular-weight PCBs such as the dioxin-like PCB 118 (Quinn et al.

2009) in the water body is likely a result of previous application in the immediate surrounding area. Typically, heavier PCB molecules are known to deposit close to the main source, resulting in relatively increased levels in the areas, even decades after initial use, whereas their lighter counterparts can travel for long distances from their sources. Although the heavier congeners

Fig. 3 Water-dissolved concentrations of some individual PCB congeners as estimated from SPMDs deployed in several sample sites

Environ Monit Assess Fig. 4 Percent composition of HCH in water of the several sample sites

were found at very low concentrations in the water phase, their high lipophilicity and biomagnification effects through the food web may be a cause of concern (Degger et al. 2011). OCPs Estimated ambient water concentrations of OCPs in the various sampling sites are shown in Table 5. Sum total OCP concentrations ranged from 0.172 ng l−1 at AUP to 37.135 ng l−1 at JR 2. The principal compositions of these compounds in all the sites were HCH and DDX (DDT, DDD and DDE). HCHs levels were higher than those of DDX with ΣHCHs ranging from 0.146 to 36.937 ng l−1 and ΣDDX varied between 0.026 and 0.549 ng l−1. Concentrations of individual isomers generally followed a similar trend as the totals. HCH concentrations from this study were in agreement with those reported by Luo et al. (2004) (5.7–23.3 ng l−1) but higher than those obtained by Wang et al. (2009) (0.10–0.63 ng l−1). Considering all the study sites, mean water-borne concentrations of individual HCH isomers increased in the order: δ-HCH (2.134 ng l−1), ε-HCH (2.250 ng l−1), γ-HCH (2.446 ng l−1), α-HCH (12.990 ng l−1) and βHCH (68.998 ng l−1), (Fig. 4), with α- and β-HCHs accounting for over 90 % of the totals. In all cases, the β-HCH predominated. This isomer is characterised by higher water solubility, lower volatility and stronger environmental stability to physical, chemical and biological degradation (Willet et al. 1998; Wang et al. 2009). Concentrations of α-HCH were also higher than for γ-, δ- and ε-isomers. Technical-grade HCH was

widely used as an insecticide, and together with its accompanying isomers, it is still readily found in the environment (Wu et al. 1997). Since this type of HCH contains between 60 and 70 % α-HCH (Li and Macdonald 2005), it is expected that for every quantity of the pesticide used (and eventually ending up in the environment), a big percentage constitutes α-HCH. Moreover, its relative volatility aids in long-range transportation to regions afar. However, the lower concentrations (than α- or β-isomers) of γ-HCH observed in all the sampling sites suggest no recent applications of the insecticide in the catchment areas of the water systems. As a signatory of the Stockholm convention, South Africa has phased out the production and use of these compounds. Sampling sites JR 1 and JR 2, both of which are found in the Jukskei River, recorded the highest water-dissolved HCH concentrations. However, a slight variation in contaminant levels between them was witnessed going downstream (from JR 1 to JR 2). Since the origin of the Jukskei River is very close to Johannesburg City, the high levels of the contaminants recorded may thus be related to previous agricultural activities. Most of the developed parts of Johannesburg long the Jukskei River were previous farms which were later sold and developed for industrial and residential properties. Further downstream (JR 2), the marginal increase in HCH levels is likely resulting from additional input from also previous agricultural activities. Site IFA recorded slightly higher water-dissolved ΣHCH concentrations than EGL, despite the two sites being in the same water body

Environ Monit Assess Fig. 5 Percent composition of DDX in water of the various sample sites

(Hartbeespoort dam). The closer proximity of the Crocodile River’s entry point into the dam to IFA than to EGL (Fig. 1) may explain this discrepancy. The summed-up concentrations of estimated free dissolved DDX in each sampling site are presented in Table 5. ΣDDX ranged from 0.026 ng l−1 in AUP to 0.549 ng l−1 in IFA. Total DDX concentrations were at most two orders of magnitude lower than HCH values. IFA and EGL recorded significantly higher contaminant concentrations (at least twofold) compared to all the other sites. DDX levels obtained in this study were comparable to those reported by Quémerals et al. (1994) and Wang et al. (2009) but lower by several orders of magnitude than those reported by Karacık et al. (2013) and Rajendran et al. (2005). Ultraviolet radiation as well as microbial activity can degrade DDT to its metabolites, DDD and DDE. These degradation products are representative of historic use of Fig. 6 PAH cross plots for the ratios Ant/(Ant + Phe) vs Flt/(Flt + Pyr)

DDT (Wang et al. 2009). The percent contributions of DDT and its metabolites in each of the sampling sites are shown in Fig. 5. Evidently, most of the DDX existed as DDD and DDE, with the former constituting the highest percentage in the majority of sites (IFA=86.8 %; JR 1= 47.4 %; JR2=62.0 %; EGL=88.3 %; HSL=38.6 %; AUP=41.7 %). This was more pronounced at IFA and EGL sampling sites. Two inferences can be made from these observations. Firstly, contamination of the sites is mainly due to past use of DDT and the contribution of current application appears limited. Secondly, reductive dechlorination mechanisms (Baxter 1990; Wedemeyer 1966) of DDT are more prevalent in the studied aquatic systems and, especially, at IFA and EGL. Sibali et al. (2008) is also reported to have looked at the level of organochlorine pesticides along the Juskei River in Johannesburg and including the Hartbeespoort Dam (HD). Soxhlet extraction was used for solid samples,

Environ Monit Assess

while water samples were extracted with liquid-liquid extraction; both techniques determines the total concentration. The concentrations of these pesticides were much higher in sediments, mostly in three-digit micrograms per kilogram levels while in water were mostly single- and double-digit micrograms per kilogram levels. High concentration in the sediment indicate accumulation from previous use. Re-desorption processes could be contributing to what is observed in water bodies.

Possible sources of PAHs PAHs enter the environment through two major pathways: pyrogenic or petrogenic sources. Water-dissolved concentrations of the analytes have been used to predict their probable sources by utilising molecular ratios of certain PAHs (Yunker et al. 2002; Zhang et al. 2004; Brandli et al. 2008; Allan and Ranneklev 2011). Specifically, variations in the ratios of the thermodynamically less-stable PAHs are used as indices in apportioning such sources. For passive samplers, Allan and Ranneklev (2011) suggest that ratios of PAHs must be for compounds with near to identical sampling rates so as to minimise bias arising from the mode of calculation of the sampling rates for compounds with widely differing log Kow values. In the current study, source apportionment of PAHs in each sampling site was attempted using ratios of Anthracene/(Anthracene + Phenanthrene) [(Ant/(Ant + Phe))] against Fluoranthene/(Fluoranthene + Pyrene) [(Flt/(Flt + Pyr))]. As shown in Fig. 6, the majority of the sites sampled gave Flt/(Flt + Pyr) ratios that were greater than 0.5, indicating pyrogenic origins. This may have occurred through combustion of biomass and coal. PAHs at sampling sites AUP and CL were clearly inclined towards petroleum combustion sources. A small ratio (