SUPPLEMENTARY INFORMATION
Investigation of 10 herbicides in surface waters of a horticultural production catchment in south-eastern Australia
Graeme Allinson a,b, AnhDuyen Bui c,a, Pei Zhang c,a, Gavin Rose c,a,
Adam M. Wightwicka,b,d, Mayumi Allinson a,b, Vincent Pettigrovea
aCentre for Aquatic Pollution Identification and Management (CAPIM), University of Melbourne, Parkville Victoria 3010, Australia
bFuture Farming Systems Research, Department of Environment and Primary Industries, 2a Bellarine Highway, Queenscliff, Victoria 3225, Australia
c Future Farming Systems Research, Department of Environment and Primary Industries, Ernest Jones Drive, Victoria 3085, Australia
dCoffey Environments, Level 1, 23 West Fyans Street, Newtown Victoria 3220, Australia
SI 1:Materials and methods
SI 1: Study area
The Yarra catchment (38°S, 145oE) is a large area (~7000 km2) that extends approximately 150 km east/north-east from Melbourne, Victoria, Australia (Figure S1). The area has undulating terrain with elevation ranging from 50 – 400 m. The soils on the northern side of the valley are grey to grey-brown loamy sands to clay loams whilst on the southern side red volcanic soils dominate. Across the catchment the soils are typically relatively acidic (pH 5.5 – 6.0) with total organic carbon contents in the range of 2 – 5 % (Wightwick et al. 2008a). The mean annual maximum temperature is ~20oC and the long-term annual rainfall is 700 - 1000 mm (BOM 2010).
Figure S1Schematic diagram illustrating the 18 monitoring sites (indicated by the circles) in the Yarra catchment, Victoria, Australia
Eighteen sites were monitored in this study, of which sixteen were in the Middle and Upper Yarra catchment area (Figure S1). The middle Yarra catchment (upstream of Yering Gorge) contains a variety of land uses, including residential, rural residential, industrial and a broad range of agricultural activities. According to MWH (2007) land use activities are production forestry (33.4%), grazing modified pastures (25.2%), rural residential (9.3%) and National Parks (6.2%). Other agricultural activities are irrigated vine fruits (2.7%), irrigated vegetable and herbs (0.6%), irrigated tree fruits (1.4%), irrigated seasonal horticulture (1.3%), irrigated flowers and bulbs (0.3%), intensive horticulture (0.8%), intensive animal production (0.8%). There are also three major sewage treatment plants that discharge in Olinda Ck, the Watts River and the Yarra River downstream of Yarra Junction. Three sites were located on the Yarra River to reflect integrated impacts and six sites were located on the lower reaches of major tributaries. Eight sites were located in the Woori Yallock catchment where a wide variety of intensive agricultural activities operate. Two sites were reference sites located in forested water supply catchments that were closed to human access. Sites were sampled in the spring and summer (September 2008 - March 2009) via the collection of spot water and surface sediment samples, as well as the deployment of time-integrated passive samplers.
SI 2.1: Water and sediment sampling
Water and sediment samples were collected from each of the 18 sampling sites every four weeks from September to December 2008 and again in February and March 2009. Water temperature, pH, and electrical conductivity (EC) were measured in-situ at the time of sampling using a field meter. The total organic carbon (TOC) content of water samples was determined according to American Public Health Association (APHA) method 5301B (APHA, 2005). Sediment samples were analysed for organic carbon (OC) using the Walkley and Black method (Rayment and Lyons, 2011, Method 6A1). The range of the mean of these water quality parameters for each of the monitoring sites was as follows: temperature (15.5 – 18.6oC; mean 16.8oC); pH (6.2 – 7.4; mean 6.7); EC (81.2 – 741 S; mean 300 S); TOC water (4.6 – 20 mg/L; mean 8.0 mg/L); sediment OC (2.0 – 7.3 %; mean 3.8 %).
Spot water samples (2 L) were collected in solvent-rinsed amber glass bottles. Sediment samples were collected by using a dip net (360 m mesh) to scoop surface sediments from the bottom of the stream/river bed. The sediment was then wet filtered on site through a 64 m mesh net into a bucket and left to settle for at least 10 minutes. To prevent cross contamination between sites the buckets and nets were thoroughly rinsed after collection and then again prior to collection at the next site. Supernatant (water) was decanted off and the wet sediment transferred into 1 Lsolvent-rinsed glass jars. After further settling in the laboratory, the remaining supernatant was decanted off, the sediment dried (40oC), and ground to <1 mm.Water and sediment samples were kept on ice during transportation to the laboratories where they were stored at 4oC prior to further processing. Water samples from the two reference sits (sites in closed water supply catchments) were used in lieu of blank water samples as part of the analytical quality control.
SI 2.2: Passive water sampling
Two types of passive sampler were used in this study, namelytrimethylpentane passive samplers (TRIMPs), a type of semi-permeable membrane device (SPMD) for non-polar chemicals with log Kow > 3.5, and the Chemcatcher™passive sampler system in a polar configuration, with rate limiting membrane, to target polar chemicals with log Kow < ~3.0. The two types of sampler were deployed for time-integrated monitoring to allow firstly a qualitative assessment (i.e. presence/absence) and then, where possible, a semi-quantitative assessment (i.e. based on estimated time-weighted average water concentrations) of herbicides in the catchment.
The TRIMPs consisted of prefabricated low density polyethylene (LDPE) membrane bags (3 cm width, 10 cm length and 40 μm mean wall thickness; Scubs Brand; Schur Consumer Products A/S, Vejle, Denmark) that were filled with 10 ml of 2,2,4-trimethylpentane (Leonard et al., 2002). Theoretically, TRIMPswill accumulate non-polar chemicals with a log Kow> 3.5, although they may also accumulate some moderately polar chemicals (i.e. log Kow 3 – 3.5; Leonard et al.,2002). In addition TRIMPsare theoretically designed to sample the ‘bioavailable’ fraction of chemicals, thus the proportion of chemical that is strongly bound to suspended particles and dissolved organic matter may not pass through the semi-permeable wall of the sampler and thus not be accumulated by the sampler.
The design of the Chemcatcher™ passive sampler (CC) is described in Kingston et al. (2000) and comprises a PTFE body, which supports a diffusion-limiting membrane and a solid receiving phase. In this study, an Empore™ SDB-XC disk (47 mm; 3M, MN, USA) was used as the receiving phase and a polyethersulfone (PES) membrane used as the diffusion-limiting membrane. The Empore™ disks and membranes were conditioned by soaking in methanol and deionised water. Once assembled the CC was filled with deionised water to cover the disk and membrane and a PTFE lid screwed on for transport.
At each site the passive samplers were deployed for sampling periods of between 21 and 28 days. The passive samplers were deployed (and retrieved) on the same days that the spot water and sediment samples were collected. No passive samplers were deployed during the period from December 2008 through to February 2009. The TRIMPs were placed inside a steel mesh cage (mesh size 12 mm) and transported to the sites in distilled water and deployed by retention of the sampler in a second, outer wire or plastic cage (all sides 15 cm, mesh size 10-12 mm). The CCs were deployed in the field by removing the transport lid, attaching the expansion ring and then suspendinga single sampler body from the top of the samemesh sampling cage containing the TRIMP sampler using a cable tie such that the receiving phase/membrane were facing downwards. At each site, duplicate cages were fixed approximately 10 cm above the stream bottom and separated by several meters. The cages were upturned under the water surface for several seconds to ensure that water completely filled the CC chamber. Upon retrieval, the solvent from each TRIMP bag was transferred to a 10 ml crimp cap glass vial with approximately 500 mg of anhydrous Na2SO4. The vial was stored at 1-4oC during transport and then storedat -20 oC at the laboratory. The CC was removed from the sampling cage, turned upside down, the expansion ring removed and the exposed chamber filled with water from the field site so that the otherwise exposed membrane was covered in water. The transport lid was then screwed on to the sampler and the CC body sealed in a plastic zip-lock bag. The CCs were also kept in an ice-cooled container during transport to the laboratory, where they were extracted within 24-48 hours. In short, the Empore™ disk and PES membrane were removed from the CC body, and dried on a hotplate (35ºC). Accumulated chemicals on the Empore™ disks were then eluted with 2 x 5 ml of methanol, with the combined eluents then evaporated to dryness under nitrogen (N2). Finally, the sample was reconstituted in 1 ml of methanol. The samples were then stored in glass sample vials in a -20C freezer in the laboratory for storage prior to analysis.
SI 2.3:Passive Sampling Quality Assurance/Quality Control
Duplicate passive samplers were deployed at each of the sites. For CCs, a field blank was included for two sites during each deployment. In short, these were CCs prepared in the same manner as those for field deployment for which at a randomly assigned site the transport lid of the field blank was removed and the water inside tipped out. The CC was then left open, exposed to the air whilst the field deployments were made, after which the sampler was re-filled with deionised water and the lid replaced. A laboratory blank was also included for each sampler retrieval and elution.
For TRIMPs, the reported values are the maximum of the duplicate determinations where both were retrieved; this conservative approach was chosen over the average as a way to represent a worst-case scenario. The agreement between duplicate TRIMPs was, however, on the whole good, with the lowest concentration observed on average within 22% of the higher concentration (minimum difference between observations, 0%; maximum 42%) For CCs, resource limitations meant that only a single disk from each site could be eluted, and so the reported values are a single residue determination.
SI 2.4:Determinationof herbicides in passive sampler, water and sediment samples
No one analytical method is appropriate for the measurement of all of the herbicides investigated in this study. Several different methods were used to prepare the different sample matrices and chemicals. For propyzamide, ananalyte most appropriately measured by gas chromatography (GC), water samples were prepared by solvent extraction. In short, an aliquot of the water sample (1 L) was shaken with dichloromethane (DCM; 100 ml), the phases allowed to separate, and the DCM decanted. This extraction procedure was repeated with 2 x 50 ml of DCM. Each of the extracts was dried through Na2SO4and combined in a flat-bottomed flask to which 3-5 drops of 1 % w/v paraffin/hexane solution had been added. A vigreaux column was added to the flat-bottomed flask and the extract concentrated on a steam bath. Finally, the solution was inverted into 2 ml of hexane. The remaining non/less volatile herbicidesinvestigated in this multi-residue study were most appropriately measured by liquid chromatography (LC). So, water samples were prepared using solid phase extraction. A Bond Elute™ PPL solid phase extraction cartridge (500mg, 3ml; Varian, Santa Clara, CA, USA) was pre-conditioned with acetonitrile and then water before being loaded with water sample (100 ml). After drying the cartridge was then eluted with acetonitrile (5 ml). The eluant was evaporated to dryness under N2 and then reconstituted in 1 ml of 50 % water/methanol before finally being passed through a 0.45 m syringe filter.
Dried sediment samples (10 g) were extracted with 35% (v/v) water/acetone (100 ml), which was subsequently filtered through a 0.45 m syringe filter and diluted with 4 % w/v Na2SO4 (650 ml). The solution was then extracted three times with DCM (100 ml, 50 ml and 50 ml). The combined DCM extract was then dried through anhydrous Na2SO4 and reduced to 5ml on a steam bath over copper filings (to remove sulphur). One millilitre of the DCM extract was inverted into 1 ml of hexane for GC analysis, while for LC analysis 1 ml of the extract was inverted into 1 ml of 50 % (v/v) methanol/water.
The 2,2,4-trimethylpentane solvent from the TRIMPs is suitable for direct injection into GC systems, and so was used directly for GC analysis, while for LC analysis 2 ml of the TRIMPs solvent was inverted into 1 ml of 50 % (v/v) methanol/water. For the CCs, which were primarily focussed on polar compounds measured using LC, each disk was eluted with methanol (5 mL) into a glass tube and the resulting solution evaporated to dryness with N2. The residue was then redissolved in 1 ml of 50 % (v/v) methanol/water for LC analysis.
The concentrations of propyzamide were determined using a Varian 3400 CX capillary gas chromatograph fitted with nitrogen-phosphorus detector (Varian, Mulgrave, Australia). An aliquot of hexane extract was simultaneously injected onto two parallel columns (15 m, 0.32 ID 0.25µm film, 95 % dimethyl -5 % diphenylpolysiloxane stationary phase (J&W™ DB-5) and a 15 m, 0.32 ID 0.25µm film, 50 % dimethyl – 50 % diphenylpolysiloxane stationary phase (J&W™DB-17)) via a split/splitless injector with a split ratio of 1:20. The GC oven was temperature-programmed (120oC 0–2 min, 120–300oC at 20oC/min, held 300oC for 1 min) for optimum separation efficiency. The injector and detector temperatures were set at 280oC and 320oC, respectively. Helium was used as carrier gas. Varian Star software (V6.0) was used to manage the chromatographic data. The organic residues were quantified by comparison with external standards. The method limit of quantification (LOQ) for propyzamide in water and sediment samples was 1.0g/L and150 g/kg (dry weight), respectively. The LOQ for TRIMPs was 250g/L. CC extracts were not screened for propyzamide.The LOQ was determined as the lowest concentration of a compound that can be reliably quantified (95 % confidence interval) in the matrix in question.
The concentrations of linuron and pendimethalin were determined using a Varian 1200L Quadrupole LC-MS/MS (Varian, Mulgrave, Australia) operating in the positive ion electrospray mode. The linuron and pendimethalin were separated from other extract components with a Varian Pursuit C18 column (150 mm x 2.0 mm) fitted with a Pursuit C18 guard column. The HPLC column was maintained at 25oC. The mobile phase consisted of (A) 20 % methanol in 5 mM ammonium acetate and (B) 90 % methanol in 5 mM ammonium acetate with the following gradient: 100 % A–100% B (0–15 min), 100 % B (15–28 min), 100 % B–100 % A (28–30 min) with a flow rate of 0.2 ml/min. Varian Workstation (V6.0) was used for data processing. Residues were quantified using external standards and each standard set was assayed a minimum of three times during each sample batch run. Sample and recovery concentrations were calculated from a linear regression of the standards. The tandem mass spectrometer was operated in the multiple reaction monitoring (MRM) mode. The method LOQ in water for linuron and pendimethalin were 0.002 and 0.01 g/L, respectively;LOQs for sediment were 1 and 5g/kg (dry weight), respectively;LOQs for TRIMPs solvent concentration were between 0.5 and 2.5g/L; CC extracts were not screened for linuron and pendimethalin.
The concentrations of the herbicides atrazine, cyanazine, hexazinone, metribuzin, prometryn, simazine, terbutryn and the atrazine metabolites desethyl atrazine (DEA) and deisopropyl atrazine (DIA) were also determined using a Varian 1200L Quadrupole LC-MS/MS (Varian, Mulgrave, Australia) operating in the positive ion electrospray mode. The triazine/triazole herbicides were separated from other extract components with a Varian C18, 5µm, Luna column (150 mm x 2.0 mm). The HPLC column was maintained at 25oC. The mobile phase consisted of (A) 20 % methanol in 5 mM ammonium acetate and (B) 90 % methanol in 5 mM ammonium acetate with the following gradient: 80% A (7 min), 70% A (12 min), 100 % B (13 min) and 100% A (7 min) at a flow rate of 0.2 ml/min. Varian Workstation (V6.0) was used for data processing. Residues were quantified using external standards and each standard set was assayed a minimum of three times during each sample batch run. Sample and recovery concentrations were calculated from a linear regression of the standards. The tandem mass spectrometer was operated in the multiple reaction monitoring (MRM) mode. The method LOQ in water for atrazine, hexazinone, and simazine were 0.01 g/L; for terbutyn and DEA, 0.02 g/L; for DIA, 0.05 g/L; for cyanazine and metribuzin, 0.1 g/L; and 0.2 g/L for prometryn, respectively. LOQs in sediment were 1g/kg (dry weight) for hexazinone, 2g/kg for atrazine and simazine, 4 g/kg for terbutryn, 16 g/kg for metribuzin, 40 g/kg for cyanazine and 45 g/kg for prometryn, respectively; LOQs for TRIMPs solvent concentration were0.5g/Lfor atrazine and hexazinone, 1 g/L for prometryn, simazine and terbutryn, 5 g/L for metribuzin, and 25 g/L for cyanazine, respectively; LOQs for CC extracts were 5g/Lfor atrazine, hexazinone, and simazine, 10 g/L for prometryn and terbutryn, 5 g/L for metribuzin, and 50 g/L for cyanazineand terbutryn, respectively.
For every analytical batch of water and sediment samples extracted, 10 % of samples were randomly selected for spike recovery determinations. Water samples were spiked with the reported herbicides in the range of 0.2.5 – 8.3 g/L. Mean recoveries ranged from75 ± 8% for atrazine (t=8.85, tcritical, p=0.05, 1T=1.943) to 98±18% (t=0.32) for hexazinone. Simazine, the most frequently reported analyte in water,recorded a mean recovery was 87±18%. In sediments, the reported herbicides were spiked in the range0.1 – 0.0.25 g/kg. The mean recoveries ranged from 72±23% (t=4.18, tcritical, p=0.05, 1T=1.796) and 76 ± 10% (t=8.52) for DIA and atrazine, respectively, to 105±14% (t=1.2) for hexazinone. Simazine, the most frequently reported analyte in sediments, recorded a mean recoveryof 94±10% (t=2.05). Pendimethalinrecovery in sediment (74±25%; t=3.67) was much better than its low recovery in water samples (16 ± 7%.). Sample results reported hereafter were not corrected for recoveries from spiked samples.
SI 3:Calculation of time-weighted average water concentrations from TRIMPs and CCs
Analysis of the TRIMPs gives a total concentration of the analyte that has accumulated in the solvent over the deployment period. Based on experimentally determined or theoretical uptake rates, the time-weighted average water concentration (CTWA) can be estimated (Leonard et al., 2002). Where log Kow was >3.5 we assumed integrative uptake of fungicides into TRIMPs and CTWAcalculated as: