Science of the Total Environment 517 (2015) 10–21

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Effects of predicted climatic changes on distribution of organic contaminants in brackish water mesocosms M. Ripszam a,⁎, C.M.J. Gallampois a, Å. Berglund b, H. Larsson c, A. Andersson b, M. Tysklind a, P. Haglund a a b c

Department of Chemistry, Umea University, 901 87 Umeå, Sweden Department of Ecology and Environmental Sciences, Umeå University, 901 87 Umeå, Sweden Umeå Marine Sciences Centre, Umeå University, Norrbyn, 905 71 Hörnefors, Sweden

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• More contaminants remained in the ecosystem at higher organic carbon levels. • More contaminants were lost in the higher temperature treatments. • The combined effects are competitive with respect to contaminant cycling. • The individual properties of each contaminant determine their respective fate.

a r t i c l e

i n f o

Article history: Received 27 November 2014 Received in revised form 10 February 2015 Accepted 13 February 2015 Available online xxxx Editor: Mark Hanson Keywords: Climate change Temperature Dissolved organic carbon Organic contaminants Environmental distribution Mesocosms

a b s t r a c t Predicted consequences of future climate change in the northern Baltic Sea include increases in sea surface temperatures and terrestrial dissolved organic carbon (DOC) runoff. These changes are expected to alter environmental distribution of anthropogenic organic contaminants (OCs). To assess likely shifts in their distributions, outdoor mesocosms were employed to mimic pelagic ecosystems at two temperatures and two DOC concentrations, current: 15 °C and 4 mg DOC L−1 and, within ranges of predicted increases, 18 °C and 6 mg DOC L−1, respectively. Selected organic contaminants were added to the mesocosms to monitor changes in their distribution induced by the treatments. OC partitioning to particulate matter and sedimentation were enhanced at the higher DOC concentration, at both temperatures, while higher losses and lower partitioning of OCs to DOC were observed at the higher temperature. No combined effects of higher temperature and DOC on partitioning were observed, possibly because of the balancing nature of these processes. Therefore, changes in OCs' fates may largely depend on whether they are most sensitive to temperature or DOC concentration rises. Bromoanilines, phenanthrene, biphenyl and naphthalene were sensitive to the rise in DOC concentration, whereas organophosphates, chlorobenzenes (PCBz) and polychlorinated biphenyls (PCBs) were more sensitive to temperature. Mitotane and diflufenican were sensitive to both temperature and DOC concentration rises individually, but not in combination. © 2015 Elsevier B.V. All rights reserved.

⁎ Corresponding author. E-mail address: [email protected] (M. Ripszam).

http://dx.doi.org/10.1016/j.scitotenv.2015.02.051 0048-9697/© 2015 Elsevier B.V. All rights reserved.

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21

1. Introduction The effects of climate change are widely studied on global and local levels. The Intergovernmental Panel on Climate Change Fifth Assessment Report (IPCC 2013) predicts that mean global temperatures will be 1–1.5 °C higher in 2016–2035 than the 1850–1900 median. Long-term predictions for the period 2081–2100 include mean global warming of 1.4–1.7 °C, with “polar amplification” (2.2 to 2.4 times the global mean increase above latitudes of 67.5° N), resulting in 3.1 to 4.8 °C increases in the northern regions (Anon., 2013). Local climate change models predict increases in sea surface temperatures, of up to 3 °C, and precipitation in the north Baltic Sea region (Neumann, 2010; Schimanke et al., 2012), accompanied by higher runoff of terrestrial dissolved organic carbon (DOC) driven by the anticipated increase in precipitation (Andersson et al., in press; Graham, 2004; Eriksson Hagg et al., 2010). Climate change is also expected to affect the environmental fates of organic contaminants (OCs), which are thought to be largely governed by temperature (Kallenborn et al., 2012; Macdonald et al., 2003). However, their fates are also influenced by numerous other variables that would be affected by anticipated climate changes, including: wind speed and direction, precipitation patterns, rising sea levels and reductions in ice cover (Kallenborn et al., 2012; Bloomfield et al., 2006; Schiedek et al., 2007; Lamon et al., 2009). Temperature increases can also enhance mobilization of OCs from reservoirs such as natural waters, soils and sediments, and alter rates of OCs' accumulation, sorption and degradation (Macdonald et al., 2003). Indirect effects of climate change on OCs may include increased riverine runoff and changes in long-range transport. Increased runoffs into the northern Baltic Sea are anticipated to result in higher DOC inflows, which may affect the amounts (and possibly profiles) of organic contaminants associated with DOC (Graham, 2004; Eriksson Hagg et al., 2010). The quantity and quality of DOC strongly influence the environmental fate of OCs in brackish water ecosystems (Krop et al., 2001; Uhle et al., 1999), such as the Baltic Sea. For example, changes in partitioning of compounds between water and DOC can affect bioaccumulation pathways in ways that depend on the structure of the food-web (Andersson, 2014). Global climate change is also expected to affect atmospheric long-range transport of OCs (Macdonald et al., 2003) and both their wet and dry deposition patterns. There have been several attempts to predict these changes using multimedia fate modeling (Kallenborn et al., 2012; Bloomfield et al., 2006; Schiedek et al., 2007; Lamon et al., 2009; Noyes et al., 2009; Dalla Valle et al., 2007). Climate change and other slow processes are difficult to study in the field. Thus, mesocosms are often used to elucidate likely responses of specific modeled ecosystems to such processes and various stressors. Notably, mesocosms have been frequently used to investigate the toxicity and physiological effects of various herbicides, insecticides, fungicides and pharmaceuticals on either single species or specific food webs (Pablo and Hyne, 2009; Nietch et al., 2013; Mohr et al., 2008; Maltby et al., 2005; Lizotte et al., 2013; Bakke, 1988). They have also been used to assess the transport and fate of dichlorodiphenyltrichloroethane (DDT), hexachlorobenzene (HCB), chlorpyrifos and PAHs (Gohil et al., 2014; Zhou et al., 2013; Pablo et al., 2008; Yamada et al., 2003), but not potential effects of climate change on the distribution and fate of OCs in (model) ecosystems. Thus, in the study presented here we investigated effects of two climatic changes (temperature and DOC rises) predicted by local climate change scenarios on the environmental distribution of OCs in Baltic Sea water mesocosms. Effects of increases in temperature (3 °C) and DOC levels (2 mg L−1) were studied both separately and in combination. These values are based on the highest predicted increase of the respective parameter (Neumann, 2010; Meier, 2006). Organic compounds were added as a mixture at sub-acute toxicity levels (0.1 μg L−1). The mixture included some legacy persistent organic pollutants (POPs), such as polycyclic aromatic hydrocarbons (PAHs), polychlorinated benzenes (PCBz), polychlorinated biphenyls (PCBs) and organochlorine

11

pesticides (OCPs), which are bioaccumulative and have well known adverse effects on living organisms including humans (Schwarzenbach, 2003; Nacci et al., 2002; Baron, 1981; Binelli and Provini, 2004). We also included emerging contaminants, including anthropogenic organophosphates (used as plasticizers and flame retardants) and biogenic low molecular weight brominated compounds (Bidleman et al., 2014). The compounds were selected to represent diverse classes of environmental contaminants, including polar and non-polar compounds and numerous functional groups. 2. Materials A DOC extract was prepared for addition to the mesocosms from soil collected from a bank of the river Öreälven located west from Norrbyn, Sweden using a spade. Its drainage basin is dominated by forested areas and is essentially undisturbed by anthropogenic activities (besides forestry and some small scale farming). The river has high concentrations of humic substances, during the spring period the river humic concentrations are N100 μg L− 1. The detailed properties of the river are described elsewhere (Andersson et al., 2013). MilliQ water was added to the soil, which consisted mainly of peat material, together with Amberlite IRC748I ion-exchange resin to remove metal ions. The resulting suspension was filtered through a 90 μm filter and the DOC content of the filtrate was determined gravimetrically following filtration through a 0.2 μm filter. Appropriate amounts of the 90 μm filtrate were added to the mesocosms (described below) to give the desired DOC concentrations. Dichloromethane, acetone, methanol, cyclohexane, toluene, and ethyl acetate (GC–MS grade) were purchased from Merck and Fischer Scientific. Whatman glass fiber filters (GF/A and Fs, 1.0 and 0.7 μm, 47 mm, 20 mm and 150 mm diameter), Eppendorf tubes and Falcon tubes were obtained from VWR International AB (Stockholm, Sweden). Spiking solutions of OCs were prepared from individual certified standard materials purchased from Dr. Ehrenstorfer GmbH (Augsburg, Germany). Isotope labeled (deuterated or 13C) internal standards were obtained from QMx laboratories Ltd. (Essex, United Kingdom) and Cambridge Isotope Laboratories Inc. (Tewksbury, MA, USA). Information about the individual OCs can be found in Tables 1 and S1 (Supplementary Information, SI). Solid-phase microextraction (SPME) assemblies including fused silica polydimethylsiloxane (FS-PDMS) fibers with a 100 μm coating mounted in a manual SPME holder were purchased from Supelco, Sigma-Aldrich AB (Stockholm, Sweden). TurboVap (Biotage EU, Uppsala, Sweden) and Heidolph (Schwabach, Germany) evaporation systems were used to concentrate samples. For size exclusion chromatography a 150 mm internal diameter Omnifit glass column (Diba Industries Ltd. Cambridge, United Kingdom) was used. It was packed in-house with 27 g of SX-3 polystyrene-divinylbenzene (PS-DVB) copolymer Bio-Beads (Bio-Rad Laboratories AB, Sundbyberg, Sweden) according to the producer's instructions (Bio-Rad Bio-Gel, 2000). 3. Methods 3.1. Experimental design Sets of brackish water mesocosms, prepared as described below, were used in a 2 × 2 full factorial experiment. The treatments were incubation at two temperatures (15 °C and 18 °C), with two DOC concentrations (ca. 4 mg L−1, the initial concentration in the brackish water, and 6 mg L−1 DOC). OCs were added to all of these systems. In addition, control mesocosms (with no added OCs) were subjected to each treatment to track changes resulting from the added compounds. All treatments were applied in triplicate, thus there 24 mesocosms in total (Fig. 1). The mesocosms were established in brand new 1 m3 polypropylene tanks (Allembalage AB, Jordbro, Sweden), which were thoroughly rinsed with high pressure hot and cold water before filling. The selected

12

Table 1 Details of the native environmental standards used throughout the experiment. Pollutant name

Biphenyl Naphthalene Fluorene Phenanthrene

Molecular formula

CAS number

C12H10 C10H8 C13H10 C14H10

Trifluralin Ethoprophos Pendimethalin Chlorpyrifos methyl Trans-chlorfenvinfos Picoxystrobin Diflufenican 4-Bromoaniline 2,4-Dibromoaniline 2.4.6-Tribromoaniline

C13H16F3N3O4 C8H19O2PS2 C13H19N3O4 C7H7Cl3NO3PS C12H14Cl3O4P C18H16F3NO4 C19H11F5N2O2 C6H6BrN C6H5Br2N C6H4Br3N

Tributyl phosphate Triphenyl phosphate Tris-(2-ClEt) phosphate Tris-(1,3-diClIp) phosphate (TDCIPP) Hexachlorobutadiene Chlorothalonil Endosulfan I Endosulfan II Alachlor 4-Bromophenol 2,4-Dibromophenol 2,4,6-Tribromophenol Pentachlorophenol

Exact concentration after 12 exchanges (ng L−1)

KOW Sw (mg L−1)

Henry's law constant (Pa ∗

Monoisotopic molecular mass (g mol−1)

Quantification ions

30 46 6.4 4.2

154.0783 128.0626 166.0783 178.0783

154.0783, 155.0816 128.0626, 129.0660 166.0783, 167.0816 178.0783, 179.0816

m3 mol−1) 25 °C

1 1 1 1

84.56 59.58 108.30 76.58

66.36 46.75 84.98 60.09

3.98 3.36 4.18 4.46

6.94 31 2 1.6

1

48.54

38.09

4.45

0.037

6.5

178.0783

178.0783, 179.0816

1 1 1 1 1 1 1 1 1 1 2 2

92.21 51.88 87.02 83.14 76.23 114.44 101.03 100.19 110.89 115.83 82.72 78.10

72.36 40.71 68.29 65.24 59.82 89.80 79.28 78.62 87.02 90.89 64.91 61.28

5 5.31 3.72 3.78 3.8 4.14 4.6 5.3 5.8 6.4 3.81 4.4

0.68 0.0047 2 2 8.52 2 6 1.8 0.085 0.009 60 0.21

37 30 1.1 0.074 1.4 1.5 21 16 27 0.67 0.011 0.22

247.8521 281.8131 287.8601

247.8521, 249.8491, 251.8492 283.8102, 285.8072, 287.8043 108.9612, 110.9582, 111.0002, 180.9379, 182.9349

188.0393 222.0003 255.9613 317.9537 304.1011 329.9020

1582-09-8 13194-48-4 40487-42-1 5598-13-0 18708-86-6 117428-22-5 83164-33-4 106-40-1 615-57-6 147-82-0

2 2 2 2 2 2 2 2 2 2

90.72 73.54 103.91 85.14 182.20 96.60 76.47 80.71 139.82 79.75

71.19 57.71 81.54 66.81 142.98 75.81 60.00 63.34 109.72 62.58

5.07 3.6 5.18 3.71 3.82 4.65 4.9 2.11 3.26 4.43

0.221 0.32 0.33 1.89 145 3.1 0.05 860 85.08 2.5

4.0 0.017 2.7 0.38 0.11 0.0006 0.0012 0.091 0.038 0.014

335.1093 242.0564 281.1376 320.8950 357.9695 367.1031 394.0741 170.9684 248.8789 326.7894

C12H27O4P C18H15O4P C6H12Cl3PO4 C9H15Cl6O4P

126-73-8 115-86-6 51805-45-9 78-43-3

2 2 2 2

82.98 89.16 78.18 123.11

65.11 69.97 61.35 96.61

4 4.59 1.51 3.65

280 1.9 950 7

0.14 0.21 0.000041 0.00026

266.1647 326.0708 283.9539 427.8839

C4Cl6

87-68-3

3

100.27

78.68

4.78

C8Cl4N2 C9H6Cl6O3S

1897-45-6 959-98-8

C14H20ClNO2 C6H5BrO C6H4Br2O C6H3Br3O C6Cl5H22O

15972-60-8 106-41-2 615-58-7 118-79-6 87-86-5

3 3 3 3 3 3 3 3

71.37 122.09 122.09 115.39 74.42 74.99 97.01 81.46

56.01 95.81 95.81 90.55 58.40 58.85 76.13 63.92

3.05 0.81 3.6 0.32 3.6 0.32 3.0 240.0 2.62 15400 2.56 2080 3.9 61.3 4.69 1000

188.0393, 190.0363 222.0003, 223.9974 255.9613, 257.9584 165.0704, 235.0081, 237.0052 152.0950, 179.1184, 304.1011 298.8836, 300.8807, 302.8777, 331.8991 264.0232, 306.0702 138.9983, 157.9625 252.0984, 253.1018 124.9826, 285.9261, 287.9231 266.9381, 268.9351, 323.0007 145.0653, 173.0603, 335.0769 266.0429, 394.0741, 170.9684, 172.9663, 248.8789, 250.8768, 252.8748 326.7894, 328.7873, 330.7853, 332.7833 98.9847, 155.0473, 211.1099 326.0708, 327.0742, 233.0368 204.9588, 63.0002 75.0002, 98.9847, 190.9432, 192.9402, 208.9537 226.8384, 224.8413, 222.8442, 259.8102 263.8816, 265.8786, 267.8757 158.9768, 159.9847, 194.9535 196.9506, 236.8413 146.0970, 160.1126, 188.1075 171.9524, 173.9503 251.8608, 249.8629, 253.8588 329.7714, 331.7693 265.8441, 263.8470

2.55

1044 0.025 1.1 1.1 0.13 0.022 0.0089 0.0036 0.0024

257.8131 263.8816 403.8169 269.1183 171.9524 251.8608 329.7714 263.8470

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21

Anthracene Phenanthrene Pentachlorobenzene Hexachlorobenzene α-HCH β-HCH Lindane (γ-HCH) δ-HCH PCB-2 PCB-11 PCB-28 Mitotane (2,4′-DDD) Diazinon Chlorthal dimethyl

92-52-4 91-20-3 86-73-7 85-01-8 85-01-8 C14H10 120-12-7 85-01-8 C6HCl5 608-93-5 118-74-1 C6Cl6 319-84-6 C6H6Cl6 319-85-7 58-89-9 319-86-8 2051-61-8 C12H9Cl C12H8Cl2 2050-67-1 7012-37-5 C12H7Cl3 53-19-0 C14H10Cl4 C12H21N2O3PS 333-41-5 1861-32-1 C10H6Cl4O4

Group Exact starting concentration (ng L−1)

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21

13

Fig. 1. Design and layout of the mesocosm experiment.

treatment temperatures were maintained using 4 swimming pools (Ultra Frame Pool 17″ (203 L) made by Intex, Long Beach, CA, USA) filled with seawater and connected to a circulating cooling system. The temperature in the tanks was monitored daily throughout the experiment. The positions of the mesocosms subjected to the different treatments were randomized to avoid systematic errors (Fig. 1). Air was continuously supplied to the mesocosms through a compressed air–valve system and bubbled up from gas lines fixed to the bottom of the tanks to ensure even oxygen supply. Mosquito nets were used to prevent insects and predators entering the mesocosms. On the 29th of May 2013, 24 mesocosm tanks were each filled with 946 L of seawater (3‰ salinity) taken from the research stations central supply of seawater, which has its intake 1 km offshore from the UMF Laboratory Norrbyn, Sweden. Care was taken to distribute the water equally between the tanks. After adding soil extracts (described below) to DC tanks (Fig. 1), the mesocosms were left to equilibrate for approximately one week before adding 2.5 mL of an OC spiking solution in methanol. The exact OC concentrations in the spiking mixture (Table 1) were measured by GC–MS using a calibration mixture prepared from the original stock solution. Throughout the experiment, 20 L of water was removed from each tank on the first, third and fifth day of each week and replaced with fresh seawater that had been passed through a 1 μm filter. The calculated dilution effect of this procedure (excluding losses) on the total OC concentration is shown in Table 1.

3.2. Weekly analyses Samples (20 L) were drawn weekly (on the same day) from the mesocosms into polypropylene canisters, using separate heatresistant rubber tubes for the sets of mesocosms subjected to each treatment. Using these samples (after 90 μm filtration), the following variables were monitored throughout the experiment: pH, chlorophyll-A, DOC, and free, particulate-bound (N0.7 μm) and total OC concentrations.

3.3. Solid phase microextraction To determine free aqueous OC concentrations, 1 L unfiltered samples were taken from the spiked mesocosms. SPME was performed using the FS-PDMS system described above, with extraction times set to 3 h at room temperature (22 °C) and constant stirring, by polytetrafluoroethylene (PTFE)-coated magnetic bars rotating at 1000 rpm. Two fibers (for which separate calibration curves were recorded) were used for each extraction to increase sample throughput. 13 C-labeled hexachlorobenzene was used as a performance standard to compensate for variations in biofouling and inconsistencies in GC–MS analysis. 3.4. Liquid–liquid extraction The fraction of OCs associated with sub-0.7 μm particles (hereafter the sub-0.7 μm fraction) was extracted by exhaustive liquid–liquid extraction, by passing 300 mL water samples through GF/F filters, followed by partitioning into dichloromethane (50 mL once and twice with 25 mL). Before the extraction, 20 μL of the internal standard (IS) mixture was added to each sample (concentrations in Table S1, SI). 3.5. Determination of KDOC values The extent of association between OCs and dissolved organic carbon (defined as particles smaller than 0.7 μm) was explored by calculating partition coefficients (KDOC in L kg− 1) for aromatic and chlorinated aromatic compounds (PAHs, PCBs, PCBz and some structurally similar pesticides), using the following equation: K DOC ¼

C LLE −C SPME −1  C DOC C SPME

where CLLE and CSPME are the total sub-0.7 μm OC concentration (ng L−1) and free contaminant concentration (ng L−1), respectively. The subtraction yields the sub-0.7 μm particle-associated concentration. CDOC is the

14

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21

DOC concentration (mg L−1). The DOC concentrations were determined by high temperature combustion and infrared spectrometry. The remaining spiked water was filtered through an active carbon filter to remove contaminants before discharge to the environment. This procedure was applied to all other samples.

the direction of air−water exchange of the target compounds from the following equation (Bidleman et al., 2014): f W CW  H ¼ fA C A  RT

3.6. Suspended particle extraction Every week (as described earlier), suspended particles in 2 L of water from each spiked mesocosm were collected by filtration on a 47 mm GF/F filter, and dried in a heater at 40 °C. The material was then extracted for analysis in 100 mL beakers by three 5 min incubations with a ternary solvent mixture of acetone, dichloromethane and methanol (1:1:1, v/v/v; 20 mL in each incubation) in an ultrasonic bath at 25 °C. The IS mixture (20 μL) was added with the first solvent fraction. The samples were then extracted in an ultrasonic bath at 25 °C for 5 min each time.

where fW and fA are the fugacities (Pa) of a selected OP in water and air, respectively. Ca is the ambient air concentration in ng m−3, and Cw is the free water concentration (ng L−1), taken as the mean of concentrations recorded on three sampling days (days 9, 16 and 24). The parameters R, T and H are the universal gas constant, the temperature in K and the compound-specific Henry's Law constant, respectively. The Henry's Law constants were obtained from published studies, or from US Environmental Protection Agency (EPA) spreadsheets (Bamford et al., 2000; Scheyer et al., 2007; Shen and Wania, 2005; Jantunen and Bidleman, 2006; Li et al., 2003). They were extrapolated to 289.63 K and are given in Pa m3 mol−1.

3.7. Size fraction separation and sample processing 3.9. GC–MS methods At the end of the experiment the top 400 L of the water in each mesocosm was fractionated into three size fractions (25, 90 and 150 μm; KC-Denmark) using nets of increasing mesh sizes fixed concentrically onto the outlet of a plastic siphon tube. 20 L sub-samples of filtrate were further filtered through a 150 mm diameter GF/F filter and a 150 mm diameter 0.2 μm cellulose acetate membrane filter. The remaining 500 L of the water and the sediment phase was only filtered using 90 and 150 μm nets, as described above. After collection the nets were washed with 0.2 μm filtered seawater and stored at −20 °C until further processing. The samples were freeze-dried and the amount of particles was determined gravimetrically before extraction with 50–60 mL of acetone–dichloromethane–methanol (1:1:1, v/v/v). All extracts were filtered through glass wool to remove particles, collected in 60 mL TurboWap tubes, evaporated almost to dryness, and dissolved in ca. 300 μL dichloromethane. After this, some sediment samples needed additional clean-up. These were dissolved in 700 μL cyclohexane–ethyl acetate (3:1 v/v) and subjected to gel permeation chromatography (GPC), using cyclohexane–ethyl acetate (3:1 v/v) as the mobile phase, with a 2 mL min−1 flow rate. The effluent was monitored using a Hewlett-Packard 1050 single wavelength UV-detector set to 254 nm. The collection windows were set using the elution times of a polystyrene (PS) mixture (1200–780 Da), bis-(2-ethylhexyl) phthalate (DEHP) and toluene. The PS mixture was used to provide reference standards for the smallest fraction of macromolecular constituents, while DEHP and toluene provided reference standards for the largest and smallest target compounds, respectively. Standards of target compounds were not used to avoid contaminating the samples. The first fraction (0–27.5 min), containing lipids and other macromolecules, was discarded; the following fraction (27.5–40 min) was collected and evaporated almost to dryness. After reconstitution in 300 μL cyclohexane the samples were ready for analysis. 3.8. Estimation of air–water exchange To elucidate the degree of volatilization of the target compounds, two polyurethane foam (PUF) integrative passive air samplers (PacWill Environmental, Beamsville, ON) were hung from the roof approximately 1 m above the mesocosms. Pairs of samplers were also placed 10 and 100 m away for comparison and tracking of other atmospheric sources. Upon completion of the experiment, the PUFs were soaked in 200 mL of dichloromethane, spiked with the IS mixture (20 μL) and extracted by ultrasonication. The extracts were then processed as described in Section 3.5. Ambient air concentrations of detected OCs (Fig. S1) were calculated from amounts detected in the PUF samples assuming an air sampling rate of 3 ± 1 m3 day−1 (Bohlin et al., 2014; Shoeib and Harner, 2002). The fugacity ratios in water and air were used to calculate

Chemical analysis was performed by GC–MS using an Agilent 7890 N GC coupled to an HRT-TOF-MS instrument manufactured by Leco (St. Joseph, MI), equipped with an ultra-inert J&W DB-5MS GC column (30 m ∗ 0.25 mm ∗ 0.25 μm) supplied by Agilent (Santa Clara, CA, USA). Samples were introduced to the GC by pulsed cold splitless injection (1 μL, splitless time 85 s) using a Gerstel cooled injection system (CIS 4) with an automatic liner-exchange (ALEX) septum head. The CIS temperature was set at 35 °C during injection, ramped at 16 °C s−1 to 300 °C and held at 300 °C for 5 min. The GC was operated in constant flow mode (1 mL min−1), with the oven temperature held at 35 °C during injection then increased at 25 °C min− 1 to 300 °C (held for 5 min). The HRT TOF-MS instrument was used in high resolution mode (R N 25000 full width at half measure). Spectra were collected in centroid mode at an acquisition rate of 3 spectra s−1, which generated enough data points across peaks for acceptable peak definition. The extraction frequency was 1.8 kHz and the mass range 38–400 m z−1. All other MS parameters and processes are specified elsewhere (Ripszam and Haglund, 2015). Analytes were quantified using the internal standard method. For the SPME extracts all detected peak areas were quantified relative to 13 C HCB. For the liquid extracts (LLE, filter and size fraction extracts) an internal standard mixture (Table S1) was used for quantification. Performance descriptors of the method, including dynamic range, regression coefficients, repeatability and instrumental limits of quantification (LOQ) are listed in Table S2. The resulting LOQs for the various sample types are listed in Table S3. Background checks were performed for all extraction methods at various stages of the analytical procedure (solvent background, solvent with concentration step, solvent with internal standard and concentration step, laboratory blanks and field blanks). Concentrations of any contaminants in samples exceeding 10% of the highest concentration measured in blanks of the same compounds were omitted from further analysis. 3.10. Data handling The average total amount of OCs (ng) in each compartment (dissolved, biota/particle size fractions and sediment) of the mesocosms subjected to each treatment was calculated, and used to calculate corresponding mass balances. For the LLE, SPME, 0.2–0.7 μm and 0.7–25 μm size fractions, the total OC amounts were calculated from the analytical data by extrapolation to the total volume of the mesocosms. Similarly, the concentrations in water and particulate size fractions were calculated by converting the measured amounts to contents per unit volume of water and unit dry weight of filtered material, respectively.

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21

In order to explore and visualize major trends in the data multivariate analysis was applied, as follows. The concentrations (ng g−1 dry weight or L−1) and amounts (ng in tanks) of OCs associated with the various fractions in each mesocosm (classified by treatment) on each sampling occasion were imported (as x values) into SIMCA (Umetrics, Umeå, Sweden) and Principal Component Analysis (PCA) was applied to explore partitioning patterns. The significance of between-treatment differences in the response variables (OC concentrations and amounts) was tested using a linear mixed-effect model in R version 3.0.2 (package lme), with temperature, DOC and the combined temperature and DOC as fixed effects and mesocosm as a random effect. Differences were regarded as significant if the p-value was less than 0.05 (Table S4). Results are presented as tSV,DF and p values, where SV and DF stand for sample variance and degree of freedom, respectively. 4. Results and discussion 4.1. Temporal changes in free water concentrations, total water concentrations, and dissolved organic carbon–water partitioning The time dependence of the free concentrations of pollutants was tracked using SPME followed by GC–MS analysis. The concentrations of most compounds rapidly declined and then stabilized at lower levels (Fig. 2). Similar trends have been reported in other environmental fate studies (Yamada et al., 2003). A plausible explanation for this pattern is that pollutants rapidly redistribute from the aqueous phase to DOC, particles, biota and the walls of the containers when mesocosms are established. Some may also be lost due to evaporation. The plateau concentration depends (inter alia) on the water solubility, stability and volatility of the OCs. The free concentrations of the more hydrophobic compounds (log KOW b 3.5, such as PCB-11, Fig. 2) generally dropped rapidly to 1–5% of the initial concentration while concentrations of the more water-soluble OCs (such as triphenyl phosphate and δ-HCH) dropped more slowly and leveled off at 15–50% of the starting concentration. Generally, the temporal variations in free aqueous concentrations did not seem to show any treatment dependency. Total dissolved concentrations of the pollutants were measured after filtration on GF/F filters, followed by LLE (Fig. 3). For the more

15

hydrophobic compounds these concentrations were higher than the free concentrations (determined by SPME) during the whole experiment. In contrast, total dissolved concentrations of many of the more hydrophilic compounds (such as organophosphates, anilines and some pesticides) correspond well with the SPME data (Fig. 2). However, there was large variability both within and between treatments in later experimental stages, likely because of larger measurement uncertainties at lower concentrations and increasing differences in microecosystem diversity with time. Partly for this reason, possibly, there were no significant differences in temporal trends among treatments. Association between the organic pollutants and DOC were studied to examine possible changes in carbon quality during the experiment, by calculating and comparing log KDOC values normalized per unit mass of DOC. The KDOC values varied considerably over time, but the only consistent temporal patterns detected during the experiment were slight increases in values for the most hydrophobic pollutants such as PCBs, PCBz and mitotane (Fig. S2). The log KDOC values also varied little between treatments, despite the approximately 2 mg L−1 difference in DOC quantity between the treatments with and without additional DOC. This indicates that there were no significant between-treatment differences in the quality of DOC, which is not unexpected as the Bothnian Sea DOC is largely of terrestrial origin (Sandberg et al., 2004). 4.2. Total mass balances and PCA The OCs were sorted into three groups based on their detection frequency and general distributions among the particulate size fractions and sediment (see Table 1). Group 1 consisted of the most hydrophobic and volatile compounds: PAHs, PCBs, PCBzs, HCH conformers and mitotane. Group 2 generally included more hydrophilic compounds, inter alia various organochlorine pesticides, organophosphates and polybromoanilines. The compounds that were infrequently detected or present at close to the blank levels (group 3) were not further analyzed and are not further discussed. Significant differences in OC distributions in the dissolved, 90 μm particulate and sediment fractions are summarized in Table S4. A mass balance of OCs in each mesocosm was created by summing amounts recorded in all compartments (Fig. 4), assuming that the OCs

Fig. 2. Changes with time (days) in free concentrations of four randomly selected organic contaminants determined by analysis of SPME extracts. The solid and dashed black lines represent concentrations under the treatments with and without additional DOC at 15 °C, respectively, while the solid and dashed gray lines represent concentrations under the treatments without and with additional DOC at 18 °C, respectively. The error bars represent the standard deviation of the concentration values within treatments (n = 3).

16

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21

Fig. 3. Changes with time in total dissolved concentrations (concentrations in the sub-0.7 μm fraction) of four randomly selected organic contaminants determined by analysis of LLE extracts (n = 3). The solid and dashed black lines represent concentrations under the treatments without and with additional DOC at 15 °C, respectively, while the solid and dashed gray lines represent concentrations under the treatments without and with additional DOC at 18 °C respectively. The error bars represent the standard deviation of the concentration values within treatments (n = 3).

were homogeneously distributed throughout the tank. This covered OCs in the sub-0.7 μm dissolved phase, the sediment, particulate matter and biota. Deficits between total recoveries and amounts added (possibly due to losses arising from volatilization and biotransformation) were considered ‘unaccounted for’. Photodegradation rates were assumed to be negligible because of the UV-filtering property of the roof above the tanks. There was a slight light intensity gradient moving from one side of the experiment to the other. This gradient did cause systematic errors because of the random positioning of the mesocosms. Quantities of OCs that diffused and sorbed to the tank walls were assumed to be the same for all mesocosms. With these assumptions, higher amounts of contaminants seem to have partitioned to biota/particles under the

high DOC concentration treatments (Fig. 4). This DOC-effect was strong at 18 °C and more moderate at 15 °C. The sedimentation was also higher in mesocosms treated with additional DOC (another effect that was stronger at 18 °C than at 15 °C). The observed temperature dependence may be attributed to higher losses of OCs at 4 mg L−1 DOC concentrations than at 6 mg L− 1 DOC concentrations. The effects of increasing the DOC level appear to counteract the effect of raising the temperature on losses, as more contaminants will be sorbed to DOC and particulate matter, thereby attenuating amounts lost due to volatilization, biodegradation. When the total OC amounts were normalized to the dry weight of samples, the previously discussed temperature and DOC effects were weaker, probably because of a dilution effect, as the total

Fig. 4. Differences among treatments in distributions of representative organic contaminants. The 15 °C and 18 °C treatments in the left and right panels, respectively; treatments without and with additional DOC in the top and bottom panels, respectively.

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21

mass of carbon in each size fraction was higher in the DOC-amended (6 mg DOC L− 1) tanks. Generally, however, concentrations of the contaminants were lower at the higher temperature (at both DOC levels) and the lower DOC level at both temperatures.

17

The PCA and significance tests revealed that the triplicate, or in some cases (due to damaged samples) duplicate, measurements of dissolved OC concentrations sometimes differed considerably (Fig. 5A and B). Unsurprisingly, no significant between-treatment differences were

Fig. 5. PCA score plots of proportions of the hydrophobic (group 1, A) and hydrophilic (group 2, B) compounds in the dissolved (sub-0.7 μm), sediment (C) N90 μm particle- and biotabound (D) amounts and sediment (E) N90 μm particle- and biota-bound (F) concentrations under the treatments: 1, 4 mg DOC L−1 at 15 °C; 2, 6 mg DOC L−1 at 15 °C; 3, 4 mg DOC L−1 at 18 °C; and 4, 6 mg DOC L−1 18 °C.

18

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21

Fig. 6. Directions of net fluxes of the detected compounds in the PUF air samplers. A value of 1 (black line) indicates that the air–water system is in equilibrium for the corresponding pollutant based on free aqueous and ambient air concentrations. Values below 1 indicate net deposition; values above 1 indicate net volatilization. The error bars show cumulative errors based on analytical and sampling rate uncertainties (n = 2).

detected by these analyses in either free (SPME extract) or total (sub-0.7 μm) dissolved contents for any OCs except biphenyl (t1,7 = 6.4, p = 0.04) and tributyl phosphate (t1,7 = 7.7, p = 0.028) (Fig. 5A and B). The aqueous concentrations showed no correlation with log KOW of the examined compounds. At the end of the experiment proportions of the more hydrophobic OCs in the sediment were higher in mesocosms containing additional DOC than in mesocosms at the same temperature lacking additional DOC (Fig. 5C). This difference was significant for almost all detected compounds with log KOW values ranging from 5.3 (hexachlorobenzene; t1,8 = 70.3, p = 0.00003) to 2.1 (4-bromoaniline; t1,7 = 8.9, p = 0.02). The amounts in sediment showed a positive but weak correlation with log KOW values (linear regression r2 values from 0.10 to 0.15).

Proportions of all monitored OCs in the sediment were also significantly lower at the higher temperature, and the strength of this temperature effect was similar for both hydrophobic compounds, e.g. PCB-2 (t1,8 = 21.4, p = 0.0017), and more polar compounds, e.g. triphenyl phosphate (t1,8 = 27.9, p = 0.0007). The results show that both the temperature and DOC concentration strongly influenced the sedimentation process, as previously noted in the mass balance evaluation. In the 0.2–0.7 μm, 25–90 μm, and 90–150 μm (Fig. 5D) and N 150 μm size fractions final proportions of contaminants were consistently higher at the higher DOC concentration. The temperature treatments had variable effects in this respect, and the between-temperature differences were greater for the hydrophobic compounds (group 1) than for the hydrophilic compounds (group 2). For many of the halogenated

Fig. 7. Effects of adding DOC on distributions of spiked 2,4-dibromoaniline (DBA) and mitotane (DDD) in the mesocosms: average percentages (with standard deviations, n = 3) of total amounts in the mesocosms with 4 mg DOC L−1 (left) and 6 mg DOC L−1 (right) at 15 °C.

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21

19

Fig. 8. Effects of adding DOC on distributions of spiked tributyl phosphate (TBP) and lindane (lin) in the mesocosms: average percentages (with standard deviations, n = 3) of total amounts in the mesocosms at 15 °C (left) and 18 °C (right) with 4 mg DOC L−1.

and aromatic compounds the differences observed were statistically significant (exemplified by the 90–150 μm fraction in Table S4). The particle sorbed fraction of the contaminants (summed up in all size fractions) showed a weak, positive correlation with log KOW values (r2 values from 0.09 to 1.61). PCA of the effects of the treatment on OC concentrations showed very similar patterns, as discussed above for the total amounts. This is illustrated by Fig. 5C, D and E, F, which respectively show differences between treatments in total amounts and proportions of OCs associated with b90 μm particles. Similar separations of groups can be seen in the two pairs of score plots (note, the orientation of objects differs in these plots, due to the difference in normalization basis). The contaminant concentrations in the sediment also showed a weak positive correlation with their log KOW values (r2 values from 0.13 to 0.16). 4.3. Air–water exchange of contaminants High amounts of legacy organic pollutants were found in the PUF samplers along with some brominated anilines and organophosphates, with detected quantities ranging from 4.1 ng to 1.6 μg per sampler. The

samplers furthest away from the mesocosms collected large amounts of naphthalene, phenanthrene, anthracene and dibutyl phosphate. These compounds were ignored in further analyses, due to high blank levels, and are not discussed further. The net fluxes (Fig. 6) were positive (from the mesocosms) and significantly different from a value of 1 (equilibrium) for biphenyl, PCBs, PCBzs and mitotane, indicating oversaturation of the water phase and net volatilization. Significant negative net fluxes were observed for TCEP, bromoanilines and diflufenican, indicating net transport from air to water. 4.4. Implications for climate change The observations of the net volatilization of some compounds from the mesocosms may be significant with regard to climate change. Hence we further evaluated the distribution of OCs at the end of each treatment (mean values for sets of three mesocosms), and observed some significant differences in sedimentation, partitioning (to biota and organic matter) and calculated losses. For example, addition of DOC to mesocosms kept at 15 °C reduced losses of mitotane (2,4′-DDD) and 2,4-dibromoaniline (DBA), as shown

Fig. 9. Combined effects of increasing DOC levels and temperature on distributions of spiked PCB-11 (P11) and phenanthrene (phe) in the mesocosms: average percentages (with standard deviations, n = 3) of total amounts in the mesocosms at 15 °C with 4 mg DOC L−1 (left) and 18 °C with 6 mg DOC L−1 (right).

20

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21

in Fig. 7, and most of the other OCs, probably by increasing their partitioning to DOC, suspended particulate matter and/or biota. Significantly increased sedimentation was found for mitotane (t1,8 = 43.9, p = 0.00017), 2,4-dibromoaniline (t1,8 = 13.9, p = 0.0058) and most of the other detected compounds. However, the aqueous concentrations did not show significant treatment-related differences. These results suggest that OCs may be more strongly retained in the future (if anticipated climate changes occur) because of higher riverine DOC influxes. Effects of increasing the temperature are exemplified by distributions of one relatively readily biodegradable compound, tributyl phosphate (TBP), and one relatively volatile compound, γ-HCH (Fig. 8). Both showed higher losses, as did all of the other OCs at the higher temperature. This could have been caused by increased volatilization and biodegradation, as both processes are directly related to temperature through their Henry's law and reaction rate constants. In summary, as climate change is expected to lead to increases in surface water temperature, reductions in the retention of organic pollutants can be expected. These findings are consistent with modeling results (Kallenborn et al., 2012; Lamon et al., 2009; Dalla Valle et al., 2007), but using empirical data we extend the simulated data, by capturing more detailed effects of climate change on brackish ecosystems, water chemistry, and distributions of contaminants. The combination of increases in temperature and DOC concentration had no consistent effects on OC distribution. Instead, the effects on the pollutants largely depended on their nature, as exemplified by the effects on phenanthrene and PCB-11 (Fig. 9). For phenanthrene there was no observable effect on dissolved amounts. Slight increases in its sedimentation and partitioning to particles were observed, but its overall retention did not change significantly. For PCB-11 combined increases in temperature and DOC concentration significantly increased losses and decreased its dissolved concentration, but not its sedimentation or partitioning to biota. This may have been due to its higher volatility compared to phenanthrene. In summary, future pollutant distribution and retention patterns are likely to be influenced by changes in both temperature and DOC concentration in the Baltic Sea. The net effect will depend on the magnitude of the change in each of these competitive factors. The change in environmental fate of each individual organic contaminant will depend on the physicochemical properties of the compound and (hence) which of two climate driven parameters influence its distribution most strongly. Acknowledgments The authors would like to thank the EcoChange project (Dnr 2009‐ 149), the Kempe Foundation, and the strong research environment “The Environment's Chemistry” at Umeå University for funding. Umeå Marine Sciences Centre (UMSC) is acknowledged for facilitating the experiment. We thank Owen Rowe for his help with setting up the experiment and Terry Bidleman for his assistance with calculating air–water exchange directions. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.scitotenv.2015.02.051. References Andersson, A., 2014. Prediction of Ecosystem Effects on the Baltic Sea at Future Climate Change Projections — Implications for Ecosystem Management. Andersson, A., Jurgensone, I., Rowe, O.F., Simonelli, P., Bignert, A., Lundberg, E., Karlsson, J., 2013. Can humic water discharge counteract eutrophication in coastal waters? PLoS One 4 (8). http://dx.doi.org/10.1371/journal.pone.0061293. Andersson, A.M., H.E.M., Ripszam, M., Rowe, O., Wikner, J., Haglund, P., Eilola, K., Legrand, C., Figueroa, D., Paczkowska, J., Lindehoff, E., Tysklind, M., Elmgren, R., 2015. Projected future climate change and Baltic Sea ecosystem management. Ambio (in press). Anon., 2013. IPCC Fifth Assessment Report. Weather 12 (68) (310–310).

Bakke, T., 1988. Physiological energetics of Littorina-littorea under combined pollutant stress in field and mesocosm studies. Mar. Ecol. Prog. Ser. 1-3 (46), 123–128. http://dx.doi.org/10.3354/meps046123. Bamford, H.A., Poster, D.L., Baker, J.E., 2000. Henry's law constants of polychlorinated biphenyl congeners and their variation with temperature. J. Chem. Eng. Data 6 (45), 1069–1074. http://dx.doi.org/10.1021/je0000266. Baron, R.L., 1981. Delayed neurotoxicity and other consequences of organo-phosphate esters. Annu. Rev. Entomol. (26), 29–48 http://dx.doi.org/10.1146/annurev.en.26. 010181.000333. Bidleman, T.F., Agosta, K., Andersson, A., Haglund, P., Nygren, O., Ripszam, M., Tysklind, M., 2014. Air–water exchange of brominated anisoles in the northern Baltic Sea. Environ. Sci. Technol. 11 (48), 6124–6132. Binelli, A., Provini, A., 2004. Risk for human health of some POPs due to fish from Lake Iseo. Ecotoxicol. Environ. Saf. 1 (58), 139–145. http://dx.doi.org/10.1016/j.ecoenv. 2003.09.014. Bio-Rad Bio-Gel, 2000. A Gels Instruction Manual 2000. http://www.bio-rad.com/ webroot/web/pdf/lsr/literature/4006139.pdf. Bloomfield, J.P., Williams, R.J., Gooddy, D.C., Cape, J.N., Guha, P., 2006. Impacts of climate change on the fate and behaviour of pesticides in surface and groundwater — a UK perspective. Sci. Total Environ. 1-3 (369), 163–177. http://dx.doi.org/10.1016/j. scitotenv.2006.05.019. Bohlin, P., Audy, O., Skrdlikova, L., Kukucka, P., Pribylova, P., Prokes, R., Vojta, S., Klanova, J., 2014. Outdoor passive air monitoring of semi volatile organic compounds (SVOCs): a critical evaluation of performance and limitations of polyurethane foam (PUF) disks. Environ. Sci. Process. Impacts 3 (16), 433–444. http://dx.doi.org/10.1039/ c3em00644a. Dalla Valle, M., Codato, E., Marcomini, A., 2007. Climate change influence on POPs distribution and fate: a case study. Chemosphere 7 (67), 1287–1295. Eriksson Hagg, H., Humborg, C., Morth, C.-M., Medina, M.R., Wulff, F., 2010. Scenario analysis on protein consumption and climate change effects on riverine N export to the Baltic Sea. Environ. Sci. Technol. 7 (44), 2379–2385. http://dx.doi.org/10.1021/ es902632p. Gohil, H., Ogram, A., Thomas, J., 2014. Stimulation of anaerobic biodegradation of DDT and its metabolites in a muck soil: laboratory microcosm and mesocosm studies. Biodegradation 5 (25), 633–642. http://dx.doi.org/10.1007/s10532-014-9687-0. Graham, L.P., 2004. Climate change effects on river flow to the Baltic Sea. Ambio 4–5 (33), 235–241. http://dx.doi.org/10.1639/0044-7447(2004)033[0235:cceorf]2.0.co;2. Jantunen, L.M., Bidleman, T.F., 2006. Henry's law constants for hexachlorobenzene, p,p′-DDE and components of technical chlordane and estimates of gas exchange for Lake Ontario. Chemosphere 10 (62), 1689–1696. http://dx.doi.org/10.1016/j. chemosphere.2005.06.035. Kallenborn, R., Halsall, C., Dellong, M., Carlsson, P., 2012. The influence of climate change on the global distribution and fate processes of anthropogenic persistent organic pollutants. J. Environ. Monit. 11 (14), 2854–2869. http://dx.doi.org/10.1039/ c2em30519d. Krop, H.B., van Noort, P.C.M., Govers, H.A.J., 2001. Determination and theoretical aspects of the equilibrium between dissolved organic matter and hydrophobic organic micropollutants in water (K-doc). Rev. Environ. Contam. Toxicol. 169 (169), 1–122. Lamon, L., Dalla Valle, M., Critto, A., Marcomini, A., 2009. Introducing an integrated climate change perspective in POPs modelling, monitoring and regulation. Environ. Pollut. 7 (157), 1971–1980. http://dx.doi.org/10.1016/j.envpol.2009.02.016. Li, N.Q., Wania, F., Lei, Y.D., Daly, G.L., 2003. A comprehensive and critical compilation, evaluation, and selection of physical–chemical property data for selected polychlorinated biphenyls. J. Phys. Chem. Ref. Data 4 (32), 1545–1590. http://dx. doi.org/10.1063/1.1562632. Lizotte, R.E., Testa, S., Locke, M.A., Steinriede, R.W., 2013. Responses of phytoplankton and Hyalella azteca to agrichemical mixtures in a constructed wetland mesocosm. Arch. Environ. Contam. Toxicol. 3 (65), 474–485. http://dx.doi.org/10.1007/s00244-0139927-z. Macdonald, R.W., Mackay, D., Li, Y.F., Hickie, B., 2003. How will global climate change affect risks from long-range transport of persistent organic pollutants? Hum. Ecol. Risk. Assess. 3 (9), 643–660. http://dx.doi.org/10.1080/713609959. Maltby, L., Blake, N., Brock, T.C.M., Van Den Brink, P.J., 2005. Insecticide species sensitivity distributions: importance of test species selection and relevance to aquatic ecosystems. Environ. Toxicol. Chem. 2 (24), 379–388. http://dx.doi.org/10.1897/04-025r.1. Meier, H.E.M., 2006. Baltic Sea climate in the late twenty-first century: a dynamical downscaling approach using two global models and two emission scenarios. Clim. Dyn. 1 (27), 39–68. http://dx.doi.org/10.1007/s00382-006-0124-x. Mohr, S., Feibicke, M., Berghahn, R., Schmiediche, R., Schmidt, R., 2008. Response of plankton communities in freshwater pond and stream mesocosms to the herbicide metazachlor. Environ. Pollut. 3 (152), 530–542. http://dx.doi.org/10.1016/j.envpol. 2007.07.010. Nacci, D.E., Kohan, M., Pelletier, M., George, E., 2002. Effects of benzo a pyrene exposure on a fish population resistant to the toxic effects of dioxin-like compounds. Aquat. Toxicol. 4 (57), 203–215. http://dx.doi.org/10.1016/s0166-445x(01)00196-5. Neumann, T., 2010. Climate-change effects on the Baltic Sea ecosystem: a model study. J. Mar. Syst. 3 (81), 213–224. Nietch, C.T., Quinlan, E.L., Lazorchak, J.M., Impellitteri, C.A., Raikow, D., Walters, D., 2013. Effects of a chronic lower range of triclosan exposure on a stream mesocosm community. Environ. Toxicol. Chem. 12 (32), 2874–2887. http://dx.doi.org/10.1002/etc.2385. Noyes, P.D., McElwee, M.K., Miller, H.D., Clark, B.W., Van Tiem, L.A., Walcott, K.C., Erwin, K.N., Levin, E.D., 2009. The toxicology of climate change: environmental contaminants in a warming world. Environ. Int. 6 (35), 971–986. Pablo, F., Hyne, R.V., 2009. Endosulfan application to a stream mesocosm: studies on fate, uptake into passive samplers and caged toxicity test with the fish M-ambigua. Arch. Environ. Contam. Toxicol. 3 (56), 525–535.

M. Ripszam et al. / Science of the Total Environment 517 (2015) 10–21 Pablo, F., Krassoi, F.R., Jones, P.R.F., Colville, A.E., Hose, G.C., Lim, R.P., 2008. Comparison of the fate and toxicity of chlorpyrifos — laboratory versus a coastal mesocosm system. Ecotoxicol. Environ. Saf. 1 (71), 219–229. http://dx.doi.org/10.1016/j.ecoenv.2007.08.022. Ripszam, M., Haglund, P., 2015. Automated method for determination of dissolved organic carbon–water distribution constants of structurally diverse pollutants using pre-equilibrium solid-phase microextraction. Environ. Toxicol. Chem. 2 (34), 266–274. http://dx.doi.org/10.1002/etc.2805. Sandberg, J., Andersson, A., Johansson, S., Wikner, J., 2004. Pelagic food web structure and carbon budget in the northern Baltic Sea: potential importance of terrigenous carbon. Mar. Ecol. Prog. Ser. (268), 13–29 http://dx.doi.org/10.3354/meps268013. Scheyer, A., Morville, S., Mirabel, P., Millet, M., 2007. Pesticides analysed in rainwater in Alsace region (Eastern France): comparison between urban and rural sites. Atmos. Environ. 34 (41), 7241–7252. http://dx.doi.org/10.1016/j.atmosenv.2007.05.025. Schiedek, D., Sundelin, B., Readman, J.W., Macdonald, R.W., 2007. Interactions between climate change and contaminants. Mar. Pollut. Bull. 12 (54), 1845–1856. http://dx. doi.org/10.1016/j.marpolbul.2007.09.020. Schimanke, S., Meier, H.E.M., Kjellstrom, E., Strandberg, G., Hordoir, R., 2012. The climate in the Baltic sea region during the last millennium simulated with a regional climate model. Clim. Past 5 (8), 1419–1433.

21

Schwarzenbach, P.R., 2003. An Introduction to Environmental Organic Chemicals in Environmental Organic Chemistry. John Wiley & Sons, Inc. Shen, L., Wania, F., 2005. Compilation, evaluation, and selection of physical–chemical property data for organochlorine pesticides. J. Chem. Eng. Data 3 (50), 742–768. http://dx.doi.org/10.1021/je049693f. Shoeib, M., Harner, T., 2002. Characterization and comparison of three passive air samplers for persistent organic pollutants. Environ. Sci. Technol. 19 (36), 4142–4151. http://dx. doi.org/10.1021/es020635t. Uhle, M.E., Chin, Y.P., Aiken, G.R., McKnight, D.M., 1999. Binding of polychlorinated biphenyls to aquatic humic substances: the role of substrate and sorbate properties on partitioning. Environ. Sci. Technol. 16 (33), 2715–2718. http://dx.doi.org/10. 1021/es9808447. Yamada, M., et al., 2003. Study on the fate of petroleum-derived polycyclic aromatic hydrocarbons (PAHs) and the effect of chemical dispersant using an enclosed ecosystem, mesocosm. Mar. Pollut. Bull. 1-6 (47), 105–113. http://dx.doi.org/10.1016/ s0025-326x(03)00102-4. Zhou, Y.Q., Tigane, T., Li, X.Z., Truu, M., Truu, J., Mander, U., 2013. Hexachlorobenzene dechlorination in constructed wetland mesocosms. Water Res. 1 (47), 102–110. http://dx.doi.org/10.1016/j.watres.2012.09.030.