Sixteen outdoor microcosms (diameter 1.8 m, total depth 0.8 m, water depth 0.5 m, water volume ca. 1,270 L) were used in the experiment. The microcosms were located at the Sinderhoeve Experimental Station (www.sinderhoeve.org) in Renkum near Wageningen, The Netherlands, and were lined with a watertight non-toxic layer of black polyethylene. Each microcosm was initially established with an 8 cm layer of sediment (fine clay) from a mesotrophic lake (dominated by the aquatic plants Elodea nuttallii and Chara sp.) and then filled with water, taken from the experimental station’s water supply basin.
In the preparatory phase, one hundred shoots of Elodea nuttallii were planted on 75% of the sediment surface of each microcosm. In addition, other macrophytes (Eleocharis acicularis, Spirodela polyrhiza, Potamogeton berchtoldii, Potamogeton pectinatus, Elodea canadensis, Potamogeton crispus and Ranunculus circinatus) developed from diaspores in the sediment during the course of study. During the pre-treatment period (3 months approximately), phytoplankton, zooplankton and macroinvertebrates were collected from uncontaminated mesotrophic ditches situated at the Sinderhoeve Experimental Station, and Veenkampen, an experimental field site of Wageningen University, Wageningen, The Netherlands and introduced into the systems in order to develop a freshwater community characteristic for lentic, edge-of-field surface water. The macroinvertebrates introduced comprised several taxonomic groups and they were representatives of various trophic levels. Dominant species included crustaceans (Asellus aquaticus, Gammarus pulex and Daphnia sp.), insects (Cloeon dipterum, Chaoborus sp., Plea minutissima, Chironomidae, odonates and trichopterans), and the non-arthropods Hirudinea (Erpobdella sp.) and Gastropoda (Valvata sp.).
During the pre-treatment period all microcosms were interconnected by tubes and the water was circulated using a pump to achieve the development of a similar biocoenoses in the test systems. The circulation of water was stopped 3 weeks before the start of the experiment.
The microcosms were investigated over a period of 7 weeks. One week prior to the first applications, all biological endpoints were sampled once to establish pre-treatment conditions, followed by a post treatment period of approximately 6 weeks.
Pesticide application and sampling
Azoxystrobin was provided by Syngenta Crop Protection AG, Switzerland as the formulated product AMISTAR® (Fluid) a 250 g a.i./L soluble concentrate formulation). There were four intended treatment regimes: (1) a continuous application treatment (CAT) of 10 μg/L (CAT10) consisting of a continuous exposure to 10 μg a.i./L for 42 days (2) a continuous application treatment of 33 μg/L (CAT33) consisting of a continuous exposure to 33 μg a.i./L for 42 days (3) a single application treatment (SAT33) consisting of a single application of 33 μg a.i./L and (4) a four application treatment (FAT16) consisting of four applications, each achieving a peak of 16 μg a.i./L with a time interval of 10 days. The treatment levels of the SAT33 and FAT16 applications were based on the 42d-TWA of 15 μg a.i./L, which fell in between the chronic exposure regimes of CAT10 and CAT33. The concentrations in the chronic tests were kept constant between 80 and 120% of desired nominal concentrations by adding more azoxystrobin during exposure. To measure the exposure concentrations, water samples from all microcosm were collected regularly (see Fig. 1). In the continuous exposure treatments, sampling and analysis of azoxystrobin were performed every 1–2 days, with dosing as necessary to maintain the concentration. Approximately 1 h after the additional application, a water sample was taken and the concentration analysed as described below. Before application, concentrations in stock and dosing solutions were checked for establishing nominal initial concentrations. The first treatment day is referred to as day 0, the first sampling as day −7 while the post first treatment days run up to day 43.
The microcosms were randomly allocated to the different treatments. All treatments were performed in triplicate with four control replicates. Azoxystrobin was applied by pouring a defined volume of dosing solution into the microcosms. The control microcosms were treated with water only. The systems were gently stirred immediately after application to promote the mixing through the water column whilst avoiding any resuspension of sediment particles and disturbance of submerged macrophytes.
Calculation of treatment level for time-variable exposures
Azoxystrobin was selected as a compound for this study, as it has a measured waterphase DT50 of 13 days in an outdoor aquatic microcosm (Jones and Lake 2000, see EFSA 2009 for a summary of this confidential report). In addition, with a log Kow of 2.5 azoxystrobin would be expected to remain mainly in the water phase (Tomlin 2011). Due to its relatively slow dissipation, exposures would be expected to be moderate to long-term. The concentrations of azoxystrobin chosen were based on the 10 μg/L NOEAEC (no observed ecologically adverse effect concentration (effect class 2/slight effects)) derived from a single application to an outdoor pond microcosm study (Cole et al. 2000). More pronounced effects may be expected when this concentration is maintained. Therefore, the intention was for the concentration in CAT10 to be equal to the 42d-TWA in SAT33 and FAT16 (calculations according to Zafar et al. (2011)). However, azoxystrobin proved to be more persistent in our microcosms and consequently the TWA concentration of the SAT33 and FAT16 were 15 μg/L instead of 10 μg/L, and therefore this TWA concentration fell in between CAT10 and CAT33.
The concentrations of azoxystrobin were determined in the water samples by taking depth-integrated water samples from the microcosms by means of stainless steel suction tubes connected to glass flasks (Schott bottle, 250 mL) using a vacuum pump. Approximately 100 mL of water were sampled from each microcosm in duplicate. Duplicate 2 mL samples from the 100 mL-water sample were transferred into 4-mL WISP vials (borosilicate) containing 2 mL of acetonitrile. The exact mass of water added was calculated by weighing the vials. The vials were closed with a cap and thoroughly shaken manually. A 2 mL high performance liquid chromatograpy (HPLC) vial was then filled with a portion of the sample, and was then sealed and analysed by Liquid chromatography-mass spectrometry with Triple Quadrupole systems (P2600 Agilent 6410 LC–MS/MS QQQ). The volume injected was 50 μL with an autosampler and the mobile phase (HPLC—water/acetonitrile; (50/50, V/V) was set at a flow rate of 1.0 mL/min. The analytical column used was an Agilent Zorbax Eclipse XDB-C18 (Diameter 4.6 mm; Length 150 mm; 5 μm). Column was set at temperature 40°C. Under these conditions, the retention time of azoxystrobin was approximately 2.40 min.
TWA concentrations of azoxystrobin were based on area under the curve (AUC) calculations, and the DT50 in SAT33 was estimated assuming first-order dissipation kinetics. Dissipation times were based on measurements for water samples above the limit of quantification (LOQ). The Level of Detection (LOD) and LOQ of the analysis were determined by adding a standard 0.01 μg/L of azoxystrobin in acetonitrile/water (v/v: 50/50) to each injection series. The concentration of this standard of 0.01 μg/L azoxystrobin was calculated from the calibration curve, while the standard itself was not part of this calibration curve. In total, this standard was injected 105 times, yielding an average concentration of 0.0208 μg/L, with a standard deviation (SD) of 0.0048 μg/L. The LOD in water sample was defined as 3 × SD (3 × 0.0048 = 0.015 μg/L), the LOQ as 10 × SD (10 × 0.0048 = 0.05 μg/L). The DT50 was calculated by means of linear regression using ln-transformed measured pesticide concentrations versus time.
Artificial substrates, consisting of litter bags (see “Decomposition” section) and pebble baskets, were used to monitor the effects of azoxystrobin on the benthic macroinvertebrate assemblage. Two pebble baskets and two litter bags were placed on concrete tiles on the sediment in each microcosm 2 weeks before the initiation of the treatments in order to allow colonisation by macroinvertebrates (for a detailed description of methods see Brock et al. (1992)).
Macroinvertebrates were sampled five times from each microcosm at days −7, 3, 10, 17 and 43. Pebble baskets were gently retrieved using a net. The litter bags were collected by hand. The substrates were first washed in a container to remove invertebrates. The macroinvertebrates were identified and counted alive, and then released back into the their original microcosms. The animals were identified to the lowest practical taxonomic level. From each microcosm abundance of macroinvertebrates from pebble baskets and litter bags were pooled prior to analysis of the data.
Phyto- and zooplankton sampling and identification
Zooplankton and phytoplankton were simultaneously sampled on days −5, 2, 9, 16, 23, 32 and 44 days by using a Perspex (Poly(methyl methacrylate)) tube (volume = 1.8 L). Depth-integrated water samples were collected from several spots in each microcosm until a bulk water sample of 12 L had been obtained in a bucket. From this bulk sample, 5 L was passed through a 55 μm mesh net to collect zooplankton. Another 5 L was passed through a 20 μm mesh net to collect phytoplankton, possibly missing the smaller phytoplankton taxa. The concentrated plankton samples were preserved with acetate buffered Lugol’s solution in a 100 mL sampling vial. The filtered water was returned into its original microcosm.
Cladocerans, copepods and ostracods (macro-zooplankton) were counted using a stereo microscope (Nikon SMZ-10, magnification 25×). Rotifers and copepod nauplii (micro-zooplankton) were quantified and identified with an inverted microscope (Carl Zeiss, Axiovert 10, magnification 100×), using a sub-sample of known volume. Rotifers and cladocerans were identified to the lowest practical taxonomic level (i.e., genus or species level), whereas copepods were identified to the suborder by classifying as calanoids or cyclopoids. A distinction was also made between nauplii and the more mature stages of the copepods.
Phytoplankton species composition was studied by counting the number of cells of a known volume which were identified to the lowest practical taxonomic level. Taxa and number of cells were based on a maximum of 200 observations, consisting of a series of 20–40 counting fields of a single cuvette under an inverted microscope (magnification 400×). Zooplankton and phytoplankton data were expressed as number of individuals per litre.
Phytoplankton chlorophyll-a was sampled in parallel with the phyto- and zooplankton sampling. One litre of the remaining from the bulk 12-l sample was used to determine the amount of chlorophyll-a of the phytoplankton. Samples were concentrated through a 1.2 μm pore size Whatmann glass-fibre filter (GF/C; diameter 4.7 cm; Maidstone, UK) using a vacuum pump. The filters containing phytoplankton were transferred into Petri dishes, wrapped in aluminium foil, and stored in a freezer at a temperature of –70°C until analysis. After ethanol extraction of the pigments, measurements of chlorophyll-a content were performed using a HPLC with fluorescence detection (Webb et al. 1992).
As an estimate of periphytic algal biomass, chlorophyll-a was sampled on day −5 and on days 2, 9, 16, 23, 32 and 42. Periphyton was sampled from glass microscope slides (7.6 × 2.6 cm) that served as artificial substrates. The slides were positioned vertically in a stainless steel frame placed in the centre of all microcosm in the north–south position tied on a long rod, approximately 10 cm below the water surface of each microcosm, and incubated for 2 weeks. The placement of frame was kept the same in all test systems during whole experimental period. On each sampling day, 8 glass slides per microcosm (colonised for 14 days) were scraped visually clean with blades (Applo Ever-Sharp-Blades; Solingen-Germany) collecting the removed periphyton in tap water. New clean slides were then reintroduced in the microcosm. The chlorophyll-a content of the water periphyton solution was analysed as described above for the phytoplankton.
Water quality parameters
Dissolved oxygen (DO), pH, electrical conductivity (EC) and temperature (T) were measured in each microcosms on days −5, 2, 9, 16, 23, 32 and 42 to detect possible changes in community metabolism. On sampling days, measurements were carried out in the morning just around the start of photoperiod, at approximately 25 cm below the water surface. h DO, pH and T were measured using a HQ40D multimeter (Hach-Lange, The Netherlands) and EC was measured with an Eijkelkamp 18.28 conductivity meter.
Alkalinity levels were determined in all microcosms prior to the initiation of the treatments (day −5) and at the end (day 43) of the experiment, using 100-mL water samples taken at a depth of 10 cm by titrating with 0.02 N HCl until a pH of 4.2 was reached (pH meter: WTW 323).
Additionally, the concentration of ammonia, nitrate, nitrite, total nitrogen, orthophosphate and total phosphate were measured in the control microcosms at the start of the experiment and in all microcosms at end of experiment (day 42). For this purpose, water samples (approximately 100 mL) were obtained from the filtered water (Whattman GF/C; 1.2 μm pore-size) collected for phytoplankton chlorophyll-a samples. These samples were transferred into 100-mL polyethylene flasks which were stored at below −18°C until analysis. Total soluble nitrogen, N–(NO2
− + NO3
+, ortho-phosphate and total phosphate were analysed using a Skalar 5100 Autoanalyser.
Decomposition of particulate organic matter (POM) was determined using litter bags (Brock et al. 1982), containing Populus × canadensis (hybrid black poplar) leaves. In the decomposition assessment, a portion of 2 g dry weight (dried at 60°C) of leaves were enclosed in each litter bag. The litter bags were made from a glass Petri-dish (diameter: 11.6 cm), closed with a cover of stainless-steel wire (mesh size: 0.7 × 0.7 mm), in which two holes (diameter: 0.5 cm) were punched to give invertebrates access to the leaves.
In each microcosm, two litter bags were placed at the sediment surface in an almost upright position for a 2-week incubation period. At the end of the incubation period, litter bags were emptied into a white tray to separate POM from adhering sediment particles and macroinvertebrates by rinsing with tap water. After sampling, a new set of litterbags was incubated. Remaining organic plant material was dried in pre-weighted aluminium foil at 105°C for 48 h to obtain dry weight. The decomposition over a 2-week period was expressed as % remaining organic material.
Macrophyte cover, biomass and bioassay
Development of macrophyte species composition and macrophyte species cover was examined three times on days −1, 14, and 44 days. Development of vegetation and the species-composition of macrophytes were investigated by monitoring macrophyte cover and abundance. The monitoring only involved the 75% of the sediment surface that was initially planted. Cover values were estimated using ordinal scales of 1 (<1%), 2 (1–5%), 3 (5–12.5%), 4 (12.5–25%), 5 (25–50%), 6 (50–75%), 7 (75–100%).
At the last sampling date (day 42), aboveground biomass of all macrophyte species were harvested for each microcosm. The plant material harvested was rinsed under tap water to remove sediment particles and macroinvertebrates and then dried in an oven in pre-weighed aluminium foil at 105°C for 48 h to determine the dry weight.
In addition to total macrophyte analysis, a Myriophyllum spicatum bioassay was performed. Flower pots (height 9.5 cm: 9 cm diameter) were filled with approximately 8.5 cm depth of sediment consisting of 86% peat, 8% sand, 6% clay and 3.73 kg fertiliser/m3 (slow release). Each pot received three apical shoots of M. spicatum with a length of 10 cm and with at least one node in the sediment. Only unbranched, non-flowering apical Myriophyllum shoots without roots were selected. In the pre-treatment period at day −21, 500 pots were introduced into one of the ditches at the Sinderhoeve Experimental Station. At day −4, 12 pots per microcosm with healthy plants were placed in plastic trays on the macrophyte-free sediment section. On day −3, 16 M. spicatum pots (one from each test cosm) were sampled to characterise the plant material (i.e., shoot and root dry weight (105°C for 24 h), shoot length and shoot number) at the time of the first application. On days 14 and 42, 6 pots per microcosm were harvested. The plants were rinsed thoroughly to remove sediment particles. The endpoints (mean per shoot) measured were aboveground dry weight, belowground dry weight (roots), total length of shoots (length of main shoot and length of side shoots), mean length of shoots (total length of shoots/total # of side shoots), and number of side shoots. For each bioassay, belowground material (roots) was separated from the aboveground parts and plant samples were dried in aluminium foil (105°C, 48 h) and weighed.
Prior to univariate and multivariate analyses, abundance data of macroinvertebrates, phytoplankton and zooplankton were ln(ax + 1) transformed, where x stands for the abundance value and ax makes 2 by taking the lowest abundance value higher than zero. We deviated from the usual ln(x + 1) transformation because the data set frequently showed low or high abundance values (i.e., 1 individual per substratum for macroinvertebrates, 0.2 individuals per litre for the zooplankton and 2 individuals per litre for the phytoplankton community). We decided that the factor ax in the ln(ax + 1) transformation should make 2 by taking the lowest abundance value higher than zero for x. A factor of two was chosen to avoid false discrepancy between zero abundance values and low abundance values. Since, for instance, the lowest abundance value higher than zero in the zooplankton data sets was 0.2, a factor 10 was used (Van den Brink et al. 2000). All other variables were tested using untransformed values. Statistically significant differences between the treatments as well as against controls were assessed for all parameters or taxon levels at each time point, using analysis of variance (ANOVA) with multiple comparison tests. ANOVA was followed by Duncan’s multiple-range test (p < 0.05), testing all treatments against the controls but also against each other. The analyses were carried out with the Genstat computer programme (v11.1, Laws Agricultural Trust, 2009 by VSN International Ltd). If the endpoint was measured more than three times after the initiation of the treatments, effects were only considered when they were consistent, i.e., occurring on least two consecutive sampling dates.
The effects of azoxystrobin treatment at the community level of macroinvertebrates, zooplankton and phytoplankton were analysed by the Principal Response Curves (PRC) method using the CANOCO software package, version 4.5 (Ter Braak and Smilauer 2002; Van den Brink and ter Braak 1999). The analysis results in a diagram showing sampling day on the x-axis and the first Principal Component of the treatment effects on the community on the y-axis (e.g., Fig. 2). The PRC method yields a diagram showing the most dominant community response to the treatment present in the data set. The species weights are shown in a separate diagram, and indicate the degree of affinity the species have with this dominant response. The results of the PRC analysis can also be evaluated in terms of the fractions of variance explained by the factors time and treatment, and the PRC diagram shows the fraction of the variance that is explained by the treatment.
In the CANOCO computer programme, redundancy analysis is accompanied by Monte Carlo permutation tests to assess the statistical significance of the effects of the treatments on the species composition of the microcosms. The significance of the PRC diagram, in terms of displayed treatment variance, was tested by Monte Carlo permutation of microcosms, using an F-type test statistic based on the eigenvalue of the component (van den Brink and ter Braak 1999). For each sampling date, all treatments were also tested against the controls using Monte Carlo permutation tests to assess the significance of treatment effects in time.