Ecological impact in ditch mesocosms of simulated spray drift from a crop protection program for potatoes

Authors


Abstract

Outdoor aquatic ditch mesocosms were treated with a range of pesticides to simulate various spray drift rates resulting from a typical crop protection program used in the cultivation of potatoes in The Netherlands. The main experimental aims of the present study were to provide information on the fate and ecological effects of drift of the pesticides into surface water and to evaluate the effectiveness of drift-reduction measures in mitigating risks. The pesticides selected and the dosage, frequency, and timing of application were based on normal agricultural practices in the potato crop. Applications of prosulfocarb, metribuzin (both herbicides), lambda-cyhalothrin (insecticide), chlorothalonil, and fluazinam (both fungicides) were made in the sequence typical of the spray calendar for potatoes. A total of 15 treatments with the various compounds were made by spray application to the water surface at 0.2%, 1%, and 5% of the recommended label rates. Chemical fate and effects on ecosystem function and structure (phytoplankton, zooplankton, chlorophyll-a, macroinvertebrates, macrophytes, breakdown of plant litter) were investigated. To interpret the observed effects, treatment concentrations were also expressed in toxic units (TU), which describe the relative toxicity of the compounds with standard toxicity test organisms (Daphnia and algae). After treatment, each compound disappeared from the water phase within 2 d, with the exception of prosulfocarb, for which 50% dissipation time (DT50) values ranged between 6 and 7 d. At the 5% treatment level, an exposure peak of 0.9 TUalgae was observed, which resulted in short-term responses of pH, oxygen, and phytoplankton. At the 5% treatment level, exposure concentrations also exceeded 0.1 TUDaphnia, and this resulted in long-term effects on zooplankton and macroinvertebrates, some of which did not fully recover by the end of the present study. At the 1% treatment level, only slight transient effects were observed on a limited number of zooplankton and macro-invertebrate species and on pH. At the 0.2% level, no consistent treatment-related effects were observed. Most of the observed effects were consistent with the results from higher-tier and mesocosm studies with the individual compounds. Multi and repeated stress played a small role within the applied pesticide package, because of rapid dissipation of most substances and the absence of many simultaneous applications. This suggests that risk assessments based on the individual compounds would in this case have been sufficiently protective for their uses in a crop protection program.

INTRODUCTION

The use of pesticides in agriculture may lead to the entry of small amounts of these compounds into aquatic ecosystems through a variety of potential routes including spray drift, drainage, surface runoff, or accidental spills (FOCUS 2001). In the flat, polder landscape of The Netherlands, spray drift into edge-of-field ditches is considered to be the main entry route of pesticides to surface water (Beltman and Adriaanse 1999). Here, these pesticides can exhibit undesirable side effects on the survival and recruitment of nontarget aquatic organisms. The entry rate of spray drift depends on the crop, application technique, implementation of risk mitigation measures, and environmental conditions on the day of application. Subsequent aquatic exposure patterns then depend on the number and frequency of applications and the dissipation rate of the compound (Boxall et al. 2002).

Regulatory schemes for pesticides are currently based on assessments of individual compounds (see EU 1997; USEPA 1998). It is common practice, however, for several different pesticides to be applied to protect crops and for their uses to be simultaneous and/or repeated. Only relatively few studies have investigated the effects of combinations of pesticides in aquatic mesocosm experiments (see Fairchild et al. 1994; Van den Brink et al. 2002; DeLorenzo and Serrano 2003), and such combinations of compounds are often not related to treatment programs used in specific crops. Only 1 previous study describes a crop-based approach to assess the combined and repeated effects of pesticides: The effects of a crop-protection program for flower bulb crops (Van Wijngaarden et al. 2004).

In this article, we describe an experiment to evaluate the fate and effects in outdoor ditch mesocosms of a typical (with respect to compounds, treatment rate, frequency, and timing) crop-protection program for potatoes in The Netherlands. Potato crops cover a large area of The Netherlands (Deneer 2003), and because of the disease pressure they face (e.g., blights such as Phytophtora infestans), are characterized by an intensive use of pesticides (see Caux, Bastien, et al. 1996). The spray drift treatment rates selected followed Dutch regulatory scenarios. The mesocosms are representative of the macrophyte-dominated drainage ditches that are typical in the Dutch agricultural landscape and harbor populations of aquatic organisms representative for several trophic levels.

MATERIAL AND METHODS

Experimental systems and study design

The experiment was performed in 12 ditch mesocosms, each with a length of 40 m, a width of 3.3 m at the water surface and 1.6 m at the sediment surface, a water depth of 0.5 m, a sediment depth of 0.25 m, and a total volume of approximately 55 m3 (Drent and Kersting 1993). Ditch sediment consisted of sandy loam with a moderate nutrient content. Construction and technical details of the mesocosms are described by Drent and Kersting (1993) (Figure 1). The systems and their aquatic community structure resemble shallow, macrophyte-dominated drainage ditches on sandy loam and clay in the agricultural landscape. The mesocosms are not dominated by free-floating Lemna species, as is predominantly the case in the hypertrophic ditches in the agricultural Dutch landscape. Therefore, they may suffer from a higher aquatic exposure to pesticides entering the systems by spray drift compared with that of Lemna-dominated ditches.

To increase similarity between ditch mesocosms (and hence increase experimental power), 2 measures were taken at the start of the experiment. First, ditches were reset to a pioneer stage by removal of the organic detritus layer in the autumn of 2000. After removal, recovery of macrophytes and macrofauna was stimulated by introduction of Elodea nuttallii shoots and macroinvertebrates originating from ditches that were situated in an area that was not in agricultural use (“Veenkampen”). Gammarus pulex was derived from the groundwater basin at the experimental station. In 2001, fine detritus and macroinvertebrates from the same area were seeded into the ditches for a 2nd time. In 2002, the year in which the experiment was performed, extra individuals of Asellus aquaticus and Proasellus juv. from ditches in the Veenkampen area were added to the ditches in the pretreatment period. Second, water was circulated between the systems for about 1 month in the pretreatment period. During the experiment, however, each ditch was isolated from the others, meaning that both duration of exposure (owing to lack of dilution) and external recovery of organisms by water (owing to lack of supply of diaspores and organisms) were more or less the worst case.

The experimental setup followed a regression design. Because spray drift is considered to be the main emission route of pesticides to surface water in the flat polder landscape of The Netherlands, we applied spray drift rates that varied from 0.2% to 5% of label-recommended rates of pesticides. Four spray drift scenarios were applied: 0%, 0.2%, 1%, and 5%. Four ditches served as control ditches. Two ditches received a low spray drift emission of 0.2%, which is the percentage that can be achieved if several mitigation measures are applied simultaneously (Ganzelmeier et al. 1995; Huijsmans et al. 1997). In 3 ditches, a spray drift emission of 1% was applied, which is the drift emission rate used for regulatory purposes for crops such as potatoes. In 3 ditches, a spray drift emission of 5% was applied, which is the spray drift emission without mitigation measures (Ganzelmeier et al. 1995; Huijsmans et al. 1997). The last scenario is meant as a realistic worst-case. The 4 treatments were randomly assigned to 1 of 3 rows of ditches at the field station (randomized block design).

Figure Figure 1..

Photo of ditch mesocosms.

The typical pesticide treatment regimes for ware and starch potatoes (Deneer 2003) were combined to develop a typical application scenario for potatoes based on active ingredients currently on the Dutch market. In potatoes, herbicides are applied at the beginning of the cultivation period, followed by various fungicide and insecticide treatments. Before harvest, aboveground crop plants are removed either mechanically or herbicidally; in this case, we assumed mechanical removal and did not apply an herbicide at the end of the season.

The pesticides used and the sequence and frequency of their application in the spray program are presented in Table 1. The use rate of the compounds was based on the label recommendations for 2003. The treatment program included prosulfocarb, metribuzin, lambda-cyhalothrin, chlorothalonil, and fluazinam (Table 1). Prosulfocarb is a thiocarbamate herbicide that inhibits weed shoot growth (Tomlin 2000). Metribuzin is a triazinone herbicide that inhibits photosynthetic electron transport and is widely used for the control of grasses and broadleaf weeds in numerous crops (Fairchild and Sappington 2002). Lambda-cyhalothrin is a synthetic pyrethroid insecticide that disrupts ion channels in nerves (Maund et al. 1998). Fluazinam is a dinitroaniline fungicide that uncouples mitochondrial oxidative phosphorilation in fungi (Tomlin 2000). Chlorothalonil is a widely used, nonsystemic chlorophenyl fungicide used to control various plant diseases in wetter regions in the United States (Caux, Kent, et al. 1996) and in The Netherlands. At least part of chlorothalonil toxicity is due to respiratory disruption (Davies 1988).

Endpoints in the ditch mesocosms (Table 2) were investigated for a period of 30 weeks (18 March–14 October 2002), including a pretreatment period of 5 weeks, a treatment period of 18 weeks, and a posttreatment period of 7 weeks. In the pretreatment period, all endpoints were sampled once or twice to establish starting conditions and to check sampling methods. In the posttreatment period, all endpoints were sampled at least twice. Measurements and sampling activities (fate sampling, artificial substrates and litter bags, macrophyte bioassays, phytoplankton and zoo-plankton sampling and measurement of water quality endpoints) were nonrandomly assigned to specific ditch segments in order to avoid mutual disturbance. On each sampling date, measurements and sampling activities were located in their assigned, specific ditch segments. Endpoints and sampling frequencies are presented in Table 2.

Pesticide application and sampling

To mimic spray drift deposition of pesticides on the water surface of the ditch mesocosms, pesticides were applied by means of a shielded spray boom (Design 1990 Technical Department of Wageningen University and Research Centre, formerly Institute of Agricultural and Environmental Engineering) (Ronday et al. 1998). By this method, the desired spray volume is dosed accurately and evenly over the water surface of each ditch and the risks of cross-contamination are minimized. The spray boom is shielded from the wind to enable its use irrespective of wind conditions. To prevent prolonged stratification of pesticide concentrations in the top layer after application, we mixed the upper water layer by moving a metal plate, which was fixed to the platforms of the movable bridges and extended to a water depth of approximately 10 to 20 cm, from one end to the other end of the ditch.

Table Table 1.. Application scheme, use rates, emission levels, calculated intended target concentrations, and nominal concentrations of pesticides (μg/L active ingredients) in the ditches
    Emission level (μ g/L)c
Active ingredientApplication weeksaUse rate (g/ha)ConcentrationbPEC 5% driftPEC 1% driftPEC 0.2% drift
  1. a Number of weeks after first pesticide (prosulfocarb) application.

  2. b Target concentrations were calculated according to the method outlined below. Nominal concentrations were based on measurements in the spray solution, the amounts of solution sprayed, and the estimated amount of water (55 m3) in the ditches. The range in nominal concentrations is given in brackets.

  3. c The following calculation method for predicted environmental concentrations (PEC) values is based on the PEC calculation for a standard ditch as described in Beltman and Adriaanse (1999). The given example is a 5% emission of prosulfocarb.

  4. Assume a ditch width of 1.0 m and a depth of 0.3 m with a ditch width at the sediment surface of 0.4 m and a side-slope angle of 1:1 (horizontal:vertical). By use of these assumptions, the ditch capacity per metre is calculated by

  5.        0.4 • 0.3 + 2 • 0.3 • 0.5 • 0.3 = 0.21 m3/m = 210 L/m

  6. where the use rate of prosulfocarb = 3,200 g active ingredient (A.I.)/ha = 0.32 g A.I./m2 = 320 mg A.I./m2 sediment surface. Assuming a 5% emission rate, 0.05 • 320 mg A.I./m2 sediment surface = 16 mg/m2 sediment surface. Therefore, 16 mg/m2 sediment surface divided by 210 L/m2 = 0.076 mg/L = 76 μg/L.

Prosulfocarb (n = 1)03200Target76153.0
   Nominal76.415.93.2
Metribuzin (n = 1)2350Target8.31.70.33
   Nominal8.21.50.27
Lambda-cyhalothrin (n = 2)5 and 95Target0.1200.0240.0048
   Nominal0.085 (0.057–0.112)0.016 (0.008–0.023)0.0040 (0.0033–0.0047)
Chlorothalonil (n = 4)6–91010Target244.80.96
   Nominal22.5 (19.8–24.0)5.1 (4.6–5.7)0.95 (0.53–1.18)
Fluazinam (n = 8)10–17200Target4.80.950.19
   Nominal4.6 (4.0–5.1)0.95 (0.9–1.1)0.18 (0.14–0.22)
Table Table 2.. Ecological endpoints and sampling frequencies
EndpointFrequency once in:
Phytoplankton
  Chlorophyll-a3 weeks
  Taxonomic analysis3 weeks
Macrophytes
  Bioassays6 weeks
  Vegetation releves6/7 weeks
Zooplankton
  Sampling (combined with phytoplankton and chlorophyll-a)3 weeks
Macroinvertebrates
  Artificial substrates4 weeks
  Litter bags4 weeks
Physicochemistry
  Oxygen, pH, temperature2–3 d
  Alkalinity and conductivity4 weeks
Nutrients
  Concentration in water3 weeks

Prosulfocarb was applied as the formulated product Boxer® (806 g/L prosulfocarb; density, 1.020 g/mL; Syngenta, UK). Metribuzin was applied as 70% metribuzin (powder; DIC 1468 70 WG, Bayer, Germany). Lambda-cyhalothrin was applied as Karate Zeon® (98 g/L lambda-cyhalothrin; density, 1.064 g/mL; Zeneca Crop Protection, UK). Chlorothalonil was applied as the formulated product Bravo® (40.4% w/w chlorothalonil; density, 1.254 g/mL; formulation YF 10934, Syngenta). Fluazinam was applied as the formulated product Shirlan® (39.5% w/w fluazinam; density, 1.273 g/mL; Syngenta). In the laboratory, a known amount of the compound was weighed into a brown glass bottle, and tapwater was added. The glass bottles were transported to the experimental site, and just before application, the entire contents of a bottle were transferred to a spraying vessel that was filled with groundwater to an end volume of 12 L, weighed, and thoroughly mixed. The ditches were sprayed for about 60 to 70 s at a pressure of 3.5 bar, resulting in a spray volume of 7 L. After the application, the spraying vessel with remaining spray solution was weighed to calculate the amount that was sprayed. Two subsamples were transferred from the remaining spray solution directly into high-pressure liquid chromatography (HPLC) vials for measuring the concentration of the application solution.

The intended treatment concentrations were calculated from product use rates, emission percentages and predicted environmental concentration (PEC) calculations for the standard Dutch ditch (Table 1) (Beltman and Adriaanse 1999). Initial concentrations in the ditches were calculated from residue measurements in the spray solution, the amounts of solution sprayed, and the estimated amount of water (55 m3) (Drent and Kersting 1993) in the ditches (Table 1). To measure exposure concentrations, water samples were collected at intervals after application (2, 8, 24, 72, 168, 336, and 672 h) in the water compartment of each ditch mesocosm or over shorter periods if dissipation was anticipated to be rapid. Depth-integrated water samples were collected with a Perspex tube (length, 50 cm; internal diameter, 3.9 cm) at 3 fixed sampling locations in each ditch. The tube was inserted vertically into the water column and was sealed with rubber stoppers. Water samples were collected from all ditches in this way. The water samples were transferred into a 1-L glass beaker and were thoroughly mixed. A subsample was then transferred into a 500-mL flask, which was sealed and transferred to the laboratory for extraction within 2 h.

Chemical analysis

Prosulfocarb, metribuzin, and fluazinam were extracted from the water samples by solid-phase extraction with OASIS extraction columns. The bottle containing the water sample was weighed before extraction. The OASIS columns (Waters® OASIS HLB extraction columns 3 mL; 60 mg) were preconditioned with 1 mL acetonitrile (LAB-SCAN HPLC quality) and 5 mL distilled water for prosulfocarb and metribuzin and with 2 mL of LAB-SCAN methanol (HPLC quality) and 5 mL distilled water for fluazinam. The bottle with the water sample was connected to the extraction column with stainless steel tubing, and water was filtered through the column. After extraction, the bottle with water sample was weighed again to measure the amount of water extracted. Prosulfocarb, metribuzin, and fluazinam were eluted from the extraction columns with acetonitrile (2 × 0.5 mL) into volumetric tubes. For each compound, both eludates were collected in a 10-mL graduated tube, and the samples were diluted with distilled water to a known final volume. A subsample was transferred into an HPLC vial, and the samples were analyzed on an HPLC system equipped with a ultraviolet detector.

Table Table 3.. Parameters for high-pressure liquid chromatography and gas chromatography/electron capture detector analysis of the pesticides prosulfocarb, metribuzin, fluazinam, lambda-cyhalothrin, and chlorothalonil
High-pressure liquid chromatography analysis
 ProsulfocarbMetribuzinFluazinam
ModelWaters M590 + Perkin Elmer ISS 100 autosampler + Waters LC 90 ultraviolet detector
Injection volume100 μL
ColumnWaters XTerraTM MSC 18, 3.5 μm, 4.6 × 150 mmWaters Novapak® C-18, 4 μm, 4.6 × 150 mm
Guard columnWaters XTerra MSC 18, 3.8 μm, 3.9 × 20 mmWaters Novapak C-18, 4 μm, 3.9 × 20 mm
Mobile phaseAcetonitrile, 65% Water, 35%Acetonitrile, 30% Water, 70%Acetonitrile, 75% Water, 25% Acetic acid, 0.1%
Flow1 mL/min
Oven temperature40°C
Wavelength (nm)220296260
Retention time (min)8.48.24.1
Detection limit (μg/L)<0.5<0.05<0.05
Gas chromatography analysis/electron capture detector analysis
 Lambda-cyhalothrinChlorothalonil
ModelHP-5890 gas chromatograph + HP 6890 autosampler
Injection volume3 μL splitless3 μL split
ColumnWCOT 25 m × 0.32 mm CP Sil 8 (film thickness, 0.4 μm)
Mobile phaseHelium 15 psi
Injection temperature240°C150°C
Temperature programInitial temperature, 80°C Initial time, 1 min Rate, 30°C/min Final temperature, 280°C Final time, 4 minInitial temperature, 120°C Initial time, 1 min Rate, 30°C/min Final temperature, 280°C Final time, 4 min
Detection temperature325°C
Retention time (min)10.35.8
Detection limit (μg/L)<0.005<0.1

Table 3 summarizes the parameters used for the HPLC analysis. Concentration calculations were based on external standard samples. Recovery efficiencies from spiked water samples (control ditches) were 92.2% (n = 15; SD 7.9%), 96.6% (n = 11; SD 3.1%), and 83.9% (n = 59; SD 3.3%) for prosulfocarb, metribuzin, and fluazinam, respectively. Concentrations were corrected for recovery.

Lambda-cyhalothrin and chlorothalonil were extracted from water samples by liquid–liquid extraction with either hexane (own distillate; lambda-cyhalothrin) or toluene (Across Organics, Belgium, toluene pro analysis; chlorothalonil). A known volume (˜300 mL) of water was transferred to a flask, and 35 mL of petroleum ether was added. The flasks were closed and shaken thoroughly for at least 15 min. The organic layer was concentrated by evaporation of the petroleum ether (own distillate) on a rotary evaporator, and the residue was dissolved in 1.5 mL hexane. Chlorothalonil samples (˜20 mL water) were extracted with 2 mL toluene by shaking thoroughly for 15 min in 35-mL tubes. Samples were analyzed without cleanup by gas chromatography with an electron capture detector. Table 3 summarizes the parameters used for the gas chromatography and electron capture detector analysis. Concentration calculations were based on external standard samples. Recovery efficiencies from spiked water samples (control ditches) were 97.5% (n = 3; SD 3.7%) for lambda-cyhalothrin and 104.9% (n = 12; SD 10.6%) for chlorothalonil. Concentrations were corrected for recovery.

The time required to dissipate 50% of the mass originally present (DT50) values were calculated according to the Organization for Economic Cooperation and Development (OECD) (2000) guideline 308 and were based on linear least-squares regression of the natural logarithm of the concentration versus time. They were calculated for mean concentrations at the 0.2%, 1%, and 5% treatment levels and were based on measurements above detection limit.

Water quality endpoints

Temperature, pH, and dissolved oxygen (DO) were measured 3 times a week between 9:00 AM and 11:00 AM. The DO was measured with a Wissenschaftlich-Technische Werkstätte (WTW) Oxi 320 Set oxygen meter at 10 and 40 cm below water surface, and pH was measured with a WTW pH196 pH meter at a water depth of 25 cm. Alkalinity and conductivity were measured every 4 weeks. Conductivity was measured by means of a WTW LF 96 conductivity meter at a depth of −25 cm. Alkalinity was analyzed in 100 mL samples, taken from a depth of 25 cm, by titration with 0.02 M hydrochloride down to pH 4.2.

Depth-integrated water samples filtered through a 40-μm mesh net (100 mL) were stored in labeled polyethylene vials in a deep freezer at a temperature below −20°C for nutrient analysis. After termination of the experiment, a Skalar 5100 Autoanalyser (Breda, The Netherlands) was used to colorimetrically analyze for nitrate/nitrite, ammonium, orthophosphate, and chloride.

Macroinvertebrates

Macroinvertebrates were sampled by means of artificial substrates and litter bags. The use of artificial substrates is described in detail by Brock et al. (1982). The litter bag technique is described in the section “decomposition”. Four weeks before the first sampling, 4 pebble baskets and 4 litter bags were placed in the ditches in order to allow colonization by microfauna and macrofauna. Pebble baskets and litter bags were assigned to 4 specific ditch segments, evenly distributed over each ditch. The pebble baskets were sampled every 4 weeks and, in total, 7 times during the experiment. Macroinvertebrates that were present on the substrates were identified and counted alive and then released again into the corresponding ditch. The species composition of macroinvertebrates was determined to the lowest practical taxonomic level. For each ditch the abundance of macroinvertebrates on pebble baskets and in litter bags were summed before analysis of the data.

Phytoplankton and zooplankton

Depth-integrated plankton samples were taken every 3 weeks. Phytoplankton and zooplankton were collected in 1 sample. Plankton samples were taken with Perspex tubes (0.4 m long, 0.8 L in volume) from 3 sampling locations assigned to specific ditch segments (˜5 samples taken at random per sampling location). Samples were mixed to create 1 sample (˜13 L volume) per experimental ditch. Five liters of each sample were filtered through a 55-μm mesh net to concentrate zooplankton and were preserved with formalin (4%). Eight liters were filtered through a 40-μm mesh net to concentrate phytoplankton and were preserved with formalin (4%). Phytoplankton samples were split up on a weight basis. One set of samples was used for identification. Of the remaining unprocessed water sample, 1 L was collected for chlorophyll-a and nutrient analysis. For the chlorophyll-a determinations, 1 L of the water sample was concentrated over a Schleicher and Schuell glassfiber filter (GF52; diameter, 4.7 cm; mesh size, 1.2 μm), by use of a vacuum pump. The filter was stored in a labelled Petri dish, wrapped in aluminium foil, at a temperature below −20°C for a maximum period of 3 months. Extraction of chlorophyll-a was performed by use of the method described by Moed and Hallegraeff (1978). Chlorophyll-a content was analyzed by spectrophotometric measurement (dichromatic following protocol NEN 6520).

Subsamples of approximately 2 to 3 mL were taken from 1 preserved sample. Subsamples were recalculated to the overall sample and the numbers in 1 L by weight. Cladocera were identified to species level. The remaining zooplankton taxa (e.g., Copepoda, Ostracoda, and Rotifera) were identified to the lowest practical level. For a few samples, it was checked whether the counting of 1 subsample was sufficient by taking an extra subsample.

Phytoplankton species composition was studied by counting the number of cells of a known volume. Taxa and number of cells were based on 40 counting fields of an object glass under a microscope (magnification, ×400). In the case of colony-forming and filamentous algae, the number of colonies/filaments was counted. Identification took place to the lowest practical taxonomic level.

Macrophytes

Seasonal development of macrophyte species composition was investigated by monitoring macrophyte cover, abundance, and structure every 6 to 7 weeks (6 times during the experiment). Cover/abundance (percentage) of each macrophyte species was estimated per 5-m ditch length by direct measurement. To describe the macrophyte structure, the mean, minimum, and maximum height of the vegetation was measured per 5-m ditch length by direct measurement.

Macrophyte bioassays were performed to determine the effects of the pesticide application regime on growth of submerged water plants. The macrophyte we used in the bioassays was Myriophyllum spicatum. Plant shoots (length, 10 cm) were derived from plants growing in the ditches. Six to 8 shoots (total weight of 5–7 g) were planted in a flowerpot. The flowerpots were filled with ditch sediment. In each ditch, 6 flowerpots were placed on the sediment surface at each of 2 locations in specific ditch segments. After an incubation period of 6 weeks, above- and below-sediment plant material was harvested, rinsed to remove sediment particles, dried in aluminium foil (105 °C, 24 h), and weighed. The material from 2 locations in each ditch was pooled. Sampling of macrophyte biomass was repeated 3 times at intervals of 6 weeks.

Leaf litter decomposition

Decomposition of coarse particulate organic material was studied by means of the litter bag technique described by Brock et al. (1982). The organic matter used in these litter bags were leaves of Populus × canadensis (2 g dry weight dried at 60 °C). These leaves had been soaked 3 times for 2 d to remove soluble humic compounds. After soaking, they were dried for 72 h at 60 °C and stored until use. We studied decomposition of these leaves as influenced by shredding macroinvertebrates. To achieve this, 2 holes of 0.5 mm were placed in the stainless steel wire (mesh size, 0.7 × 0.7 mm) around the glass Petri dishes (diameter, 11.6 cm) in order to allow macroinvertebrates to enter. Two litter bags were introduced at the sediment surface of a ditch at each of 2 locations. After an incubation period of 2 weeks, each litter bag was emptied in a white tray to separate coarse particulate organic material and macroinvertebrates. Macroinvertebrates on the outer surface of the litter bags (snails, leeches) were removed and not included in the sample. Macroinvertebrates were identified alive, counted, and released again into the ditch. Remaining organic plant material was dried in aluminium foil at a temperature of 105 °C. After 24 h, dry weight was determined. The decomposition over a 2-week period was expressed as percentage of remaining organic material. New litter bags were incubated 2 weeks before the next sampling date. Decomposition measurements were repeated 7 times.

Toxic units

To assist with the interpretation of the observed effects, we calculated Toxic Units (TU) and applied the concept of concentration addition (Deneer 2000). The TU were based on acute toxicity data of the most sensitive standard test species and measured exposure concentrations in the water compartment of the ditches. The toxicity for sensitive invertebrates (zooplankton and macroinvertebrates) was scaled to the toxicity of the 5 pesticides for Daphnia magna:

equation image((1))

where Ci is the actual concentration of the compound i, EC50i is the geometric mean 48-h EC50 of compound i for D. magna (Table 4), and y is the resulting fraction of TU. The toxicity for phytoplankton was scaled to the toxicity of the 5 pesticides for algae by using Equation 1. EC50i is the geometric mean EC50 for Selenastrum capricornutum.

The toxicity of the pesticide package for macrophytes was calculated on the basis of the toxicity of the herbicides prosulfocarb, metribuzin, and chlorothalonil to Lemna minor or L. gibba. No data were available for either of the other compounds. Previous studies with pesticides in experimental ecosystems have demonstrated that effects on primary producers are likely to occur at TUalgae > 0.1, and effects on invertebrates are likely to occur at TUDaphnia > 0.01–0.1 (Brock, Lahr, et al. 2000; Brock, Van Wijngaarden, et al. 2000).

Data analysis

In the data analysis, we used a combination of multivariate (principal response curves [PRC]) and univariate (Williams test) techniques. The PRC technique serves to identify the treatment levels that affect the community and to indicate the taxa for which the responses affiliate most to the treatment regime. The Williams test is used to analyze the dynamics at the taxon level and helps to further identify the number of taxa affected. A graphical interpretation of the univariate analysis is often needed to decide whether the statistically significant deviations can be considered consistent effects and whether these deviations make any sense in relation to variations in abundance. At the taxon level, no observed effect concentrations (NOEC) resulted from the Williams test (ANOVA, p ≤ 0.05) (Williams 1972). Analyses were made with the Community Analysis computer program (Hommen et al. 1994). Data of macrophyte bioassays and leaf litter composition were also analysed with the Williams test (ANOVA, p ≤ 0.05) (Williams 1972).

Table Table 4.. Geometric means of toxicity data of species representative of primary producers and invertebrates for the pesticides used in the experimenta
PesticideTaxonEC50 (μg/L)NrExposure time
  1. a Geometric mean effect concentrations (EC50s) (μg/L) are based on toxicity data for standard laboratory test species (Nr = number of values) commonly used in the 1st-tier risk assessment procedure for the administration of pesticides. Data were from the ECOTOX data base (www.epa.gov./ecotox/), from the RIVM database (Rijksinstituut voor Volksgezondheid en Milieu, Bilthoven, The Netherlands), and from data kindly provided by industry (Syngenta, Jealotts Hill, UK).

  2. b Data originate from Brock et al. (2004).

  3. c Data originate from Van Wijngaarden et al. (2004).

ProsulfocarbSelenastrum capricornutum86.7196 h
 Lemna gibba690114 d
 Daphnia magna1531248 h
MetribuzinbS. capricornutum39.7 72–96 h
 L. minor36.5 96 h
 D. magna14,983 48 h
Lambda-cyhalothrincS. capricornutum>1000 96 h
 Lemna sp. 
 D. magna0.35 48 h
 Oncorhynchus mykiss0.32 96 h
ChlorothalonilS. capricornutum190.4372–120 h
 L. gibba510114 d
 D. magna105.2748 h
FluazinamS. capricornutum160196 h
 L. gibba 
 D. magna201.9448 h

Before univariate and multivariate analysis of abundance values, the macroinvertebrate data were ln(2x + 1) transformed, the zooplankton data were ln(10x + 1) transformed, and the phytoplankton data were ln(0.2x + 1) transformed, where x equals the abundance value. This transformation was performed to down-weight high abundance values and approximate a normal distribution of the data (for rationale, see Van den Brink et al. 2000).

The effects of the treatment on the community level of macroinvertebrates and zooplankton were analyzed by the PRC method, which is based on the redundancy analysis ordination technique, the constrained form of principal component analysis (Van den Brink and Ter Braak 1998, 1999). The PRC method is a multivariate technique specially designed for the analysis of data from microcosm and mesocosm experiments. PRC produces a diagram that shows the sampling weeks on the x-axis and the 1st principal component of the treatment effects on the y-axis. This yields a diagram that shows the deviations in time of the treatments compared with the control. The species weights (bk), shown on the right side of the diagram, can be interpreted as a correlation of each species with the response given in the diagram. Thus, the species that has the highest weight is indicated to have decreased most at the highest treatment level. The species with the lowest negative weight has increased most at the highest treatment level. A complete description and discussion of the PRC method is given by Van den Brink and Ter Braak (1998, 1999). PRC analysis was performed by means of Canoco for Windows software package, version 4 (Ter Braak and Smilauer 1998).

In the Canoco computer program, redundancy analysis is accompanied by Monte Carlo permutation tests to assess the statistical significance of the effects of the explanatory variables on species composition of the samples (Van den Brink et al. 1996). The significance of the PRC diagram in terms of displayed treatment variance was tested by Monte Carlo permutation of entire time series in the redundancy analysis from which PRC is obtained, by using an F-type test statistic based on the eigenvalue of the component (Van den Brink and Ter Braak 1999).

Although the 1st principal component extracts the maximum amount of information from the multivariate treatment effects, it does not necessarily describe the effects of each treatment on all taxa in sufficient detail. Further components can be extracted from the residual variation. The significance of the 2nd and higher PRC diagrams was tested by means of Monte Carlo permutation tests. It showed that the 2nd and higher components were not significant and were therefore not considered. The results of the PRC analysis can be evaluated in terms of the fractions of the variance explained by the factors time and treatment. Which fraction of the variance is explained by treatment, is shown in the PRC diagram.

Monte Carlo permutation tests were also performed for each sampling date, by use of the ln-transformed treatment levels as the explanatory variable (for rationale, see Van den Brink et al. 1996). This procedure served to test the significance of the treatment for each sampling date. In addition to the overall significance of the treatment regime, we tested which treatments differed significantly from the controls, so as to infer the NOEC at the community level (NOECcommunity). The NOECcommunity calculations were done by applying the Williams test to the sample scores of the 1st principal component of the PCA of each sampling date in turn (for the rationale of this, see Van den Brink et al. 1996).

RESULTS

Exposure concentrations

Overall, initial treatment concentrations (based on measurements in the spray solution and the amounts of solution sprayed) in the ditch mesocosms were close to the intended peak concentrations (Table 1). The only exception was that the 2nd application of lambda-cyhalothrin was approximately 50% lower than intended, which could be ascribed to an incorrect 50% lower concentration in the application solution. Average treatment concentrations in the treatment regimes as a function of time are shown in Figure 2.

Prosulfocarb disappeared from the water layer most slowly. Mean calculated dissipation DT50 value from the water layer was about 7 d at the 5% treatment level (Table 5). Disspation DT50 values for metribuzin and fluazinam in the water compartment were between 1 and 2 d (Table 5). For lambda-cyhalothrin, a DT50 value in the water layer of about 1 d was calculated. With water DT50 values of 0.3 to 0.5 d, dissipation of chlorothalonil was the most rapid (Table 5).

Peak exposure concentrations reached 0.88 TU for algae at the highest treatment level (5%) (Figure 3). The largest component of the toxicity was caused by prosulfocarb in the 1st weeks of the experiment. Based on toxic units, metribuzin was expected to contribute to the toxicity for algae as well (˜0.1 TU; Figure 3). However, data from mesocosm studies indicate that effects of metribuzin at the ecosystem level are unlikely to occur at the applied concentrations. The contribution of chlorothalonil to the toxicity to algae was low (short exposures to 0.12–0.13 TU), whereas the contribution of fluazinam was negligible. At the 1% treatment level, only prosulfocarb reached concentrations at which toxicity for algae could be expected (Figure 3).

For invertebrates, peak exposure concentrations reached 0.4 TU at the highest treatment level (5%) (Figure 3). For these groups, lambda-cyhalothrin and chlorothalonil were expected to contribute most to the observed toxicity. Contributions of the other compounds at the 5% treatment level and of lambda-cyhalothrin and chlorothalonil at the 1% treatment level lie in the range 0.01 to 0.1 TU. In this range, low levels of effects cannot be ruled out. At the 0.2% level, TU values were less than 0.014 for invertebrates and less than 0.06 for algae, so no effects were anticipated.

Water quality endpoints

Nutrient concentrations in the water layer were low (Table 6). Nutrients and chloride were not different between the treatments.

Table 7 presents the results of NOEC calculations for pH and DO. There was a significant decrease in pH of approximately 0.5 to 1 pH unit from day 1 to day 16 at the 5% treatment level, and from day 1 to day 9 at the 1% treatment level. In addition, there was a significant decrease of oxygen of approximately 0.75 mg/L on day 3 during this period at the 5% treatment level. The responses observed can most probably be attributed to prosulfocarb. A significant decrease in pH also occurred from day 49 to day 53 at the 5% treatment level, overlapping slightly with the application period of chlorothalonil. In the last week of the application period (days 112 through 123), DO in the water column significantly increased at the 5% treatment level (Table 7). Increased oxygen levels coincided with an increase in turbidity of the water at the 1% and 5% treatment levels (data not shown), and an increase in chlorophyll-a (NOEC week 14 = 0.2%), indicating an increase in algal densities.

Figure Figure 2..

Trends in concentrations of pesticides applied in the potato spray program. Concentrations have been measured in the water compartment of the ditch mesocosms in the 0.2%, 1%, and 5% treatment regimes. Note that the different pesticides were applied on different dates (see Table 1).

Macroinvertebrates

Over the experimental period, a total of 81 different macroinvertebrate taxa were identified as belonging to 8 classes, with almost half of the taxa (34) being Insecta. The mesocosms were characterized by Crustacea (dominant species: G. pulex, Proasellus meridianus/coxalis), Insecta (Ephemeroptera; dominant species: Cloeon dipterum, Caenis horaria; Odonata; Diptera: Chaoborus sp., Tanypodinae, Ceratopogonidae), Hirudinea (dominant species: Erpobdella octoculata), Gastropoda (dominant species: Armiger cristata) and Turbellaria (dominant species: Polycelis nigra/tenuis).

Table Table 5.. Calculated dissipation DT50 (dissipation time, time required to dissipate 50% of the mass originally present) values (d) for the compounds in the pesticide package in the water column of the ditchesa
 ProsulfocarbMetribuzinChlorothalonilLambda-cyhalothrinFluazinam
Drift (%)MedLowHighMedLowHighMedLowHighMedLowHighMedLowHigh
  1. a The table presents median values over time for the 3 different treatments, as well as 95% lowest and highest values. NC = could not be calculated because of very fast dissipation of the compound; med = median; low = lowest value; high = highest value.

0.26.15.66.71.31.11.60.30.11.0NCNCNC2.01.110.0
16.05.46.71.71.32.60.30.12.71.20.52.71.71.22.9
57.06.67.51.71.42.20.50.31.20.90.71.21.41.31.6

Of the total variance, 24% could be attributed to the treatment regime by the PRC analysis (Figure 4). The PRC diagram displayed a significant amount of the treatment variance (p = 0.003). Clear treatment effects were observed at the 5% and 1% treatment levels compared with the controls from week 7 onward (Figure 4 and Table 8). At the 5% level, the PRC diagram shows long-lasting effects on the macro-invertebrate community without full recovery (Figure 4). These effects first became evident after the 1st application of lambda-cyhalothrin. Effects at the community level also persisted during the application periods of chlorothalonil and fluazinam, presumably because there was no recovery from the lambda-cyhalothrin effects. At the 1% level, effects at the community level were 1st observed in week 7 and seemed to be associated with the application of lambda-cyhalothrin in week 5. A similar effect was not observed after the 2nd application of lambda-cyhalothrin, possibly because dosages were lower than the intended peak concentration and just below or above threshold levels (Figure 2 and Table 11).

Statistical calculations resulted in an isolated community level NOECmacroinvertebrates at the 0.2% treatment level and an overall NOECcommunity of 1% (Table 8).

Taxa with high positive weights (i.e., those taxa that decreased most strongly in association with changes in community structure) in the PRC analysis (Figure 4) were G. pulex (Figure 5A), Chaoborus sp. (Figure 5B), C. horaria (Figure 5C), and C. dipterum (Figure 5D). The populations of these taxa declined significantly at the highest treatment level. Recovery was observed for the affected insect taxa but was not observed for G. pulex. Taxa with high negative weights shown in the PRC diagram (Figure 4) comprise Dero sp. (Figure 5F), Stylaria lacustris (Figure 5I), and Orthocladiinae/Chironomini (Figure 5H). The population abundance of these taxa all increased, at least temporarily.

Figure Figure 3..

Toxicity of the pesticide package for invertebrates and algae at 1% and 5% drift emission of label-recommended rates of pesticides expressed as Toxic Units (TU) and the contribution of each pesticide to the toxicity (0.2% not shown). See text for rationale for these values. The lines of 0.01 and 0.1 TU are marked.

Table Table 6.. Concentrations of nutrients and chloride (mg/L) in surface water of ditch mesocosms
 Sum nitrate/nitriteAmmoniumOrthophosphateChloride
Controls0.079 ± 0.0890.032 ± 0.0120.006 ± 0.0026.656 ± 0.740
0.2%0.081 ± 0.0740.036 ± 0.0110.006 ± 0.0027.026 ± 0.890
1%0.094 ± 0.1140.029 ± 0.0100.007 ± 0.0037.124 ± 0.819
5%0.082 ± 0.1040.029 ± 0.0140.006 ± 0.0026.977 ± 0.737

Statistical analysis of treatment-related responses for individual macroinvertebrate populations resulted in NOECs for 8 taxa (Table 9). Adverse effects of treatments were apparent from week 7 onward. After week 15, recovery of most insect populations has occurred. Consistent responses, defined as statistically significant deviations pointing in the same direction on at least 2 consecutive sampling dates, were observed for all the presented taxa in Table 9. Longer-lasting reductions were observed at the 5% treatment levels and occurred with G. pulex (Figure 5A). Recovery of the populations of this species was hindered by isolation of the ditches. At the 1% treatment level, long-lasting reductions were observed within Chaoborus sp. (Figure 5B) and C. horaria (Figure 5C). The lowest consistent NOECs were calculated for Chaoborus sp. at the 0.2% treatment level. For C. horaria and Orthocladiinae/Chironomini, an isolated NOEC was calculated of 0.2%. NOECs for other species and sampling dates were at the 1% treatment level.

Phytoplankton and zooplankton

Over the experimental period, a total of 59 different zooplankton taxa were identified. The majority of the taxa (40) belonged to the Rotifera, followed by Cladocera (13), Copepoda (3), Ostracoda (1), and Diptera (3). Copepoda and Ostracoda were not identified to species level. Rotifera were the most abundant.

Of the total variance, 18% could be attributed to the treatment regime by the PRC analysis (Figure 6). This percentage is in line with results from other experimental ponds (see Sanderson 2002). The PRC diagram displayed a significant amount of the treatment variance using a threshold value of 0.10 for p (p = 0.068). It shows a clear trend, particularly for the 5% treatment level (Figure 6). Monte Carlo permutation tests showed significant reductions at this level from week 8 through week 14 after application (Table 8). The 1st effects on zooplankton followed the application of the insecticide lambda-cyhalothrin. The period in which effects on zooplankton were apparent also overlapped with the application period of the fungicide chlorothalonil. Recovery had already started within the application period of the fungicide fluazinam. Statistical calculations resulted in a NOECcommunity for the zooplankton at the 1% treatment level (Table 8).

A number of Rotifera, including Polyarthra remata and Hexarthra sp., showed high negative weights in the PRC (Fig 6; see species weights bk), indicating a treatment-related decline. Cladocera had positive weights in the PRC, indicating treatment-related increases in abundance. Treatment-related responses for individual taxa were statistically analyzed and resulted in a range of NOECs (Williams test, p ≤ 0.05) for 3 zooplankton taxa (Table 9). For Hexarthra sp. and Daphnia group galeata, a NOEC of 0.2% was calculated on an isolated sampling date as well as a NOEC of 1% on several other sampling dates. For P. remata a NOEC of 1% was calculated.

The dynamics of zooplankton populations showing treatment-related responses after application of the pesticide package are presented in Figure 7. Significant long-lasting reductions in numbers occurred in 2 populations of Rotifera species. In the population of Hexarthra sp., reductions were observed at the 1% and 5% levels (Figure 7A). In the population of P. remata, reductions were observed at the 5% level (Figure 7B). Within the group of Calanoida, reductions were observed at the 5% level (Figure 7D). Short-term increases were observed in the population of Daphnia group galeata at the 1% and 5% levels (Figure 7C). Both positive and negative effects were apparent from week 8 onward.

A total of 92 phytoplankton taxa were identified. Dominant groups were green algae, mainly Chlorococcales, Desmidiaceae, and Diatoms. The PRC diagram for the phytoplankton did not display a significant amount of the treatment variance and, thus, is not shown. Statistical calculations resulted in a NOECcommunity for the phytoplankton at the 1% treatment level (Table 8), which was calculated for 1 sampling date (week 5). The data also suggest a trend of an effect in week 2, although not statistically significant (p = 0.08; Table 8). Of the phytoplankton taxa, only Flagellatae showed a consistent response (increase in numbers) at the 5% treatment level in weeks 2 and 5 (Table 9). For Flagellatae, a NOEC of 1% was calculated. Figure 8 presents the dynamics of the population of Flagellatae.

Chlorophyll-a concentrations of phytoplankton did not show a consistent treatment-related response. Only 1 isolated NOEC could be calculated for week 14 (0.2%).

Macrophytes

The aquatic vegetation in the ditches was dominated by Chara globularis ssp. virgata. Other abundant macrophytes were Elodea nuttallii and Sagittaria sagittifolia. Submerged and floating algal beds were locally abundant.

Of the percentage cover/abundance data of macrophyte taxa, only floating filamentous algae showed a consistent response (Table 9) after the application period of the pesticide package. Floating filamentous algae developed in controls and 0.2% treatments and resulted in significantly higher percentage cover compared with that of (Table 10) the 1% and 5% treatments. Dynamics of submerged filamentous algae showed a consistent treatment-related response as well (Table 10). There were no significant treatment-related responses in 4 in situ bioassays performed with M. spicatum (NOECs > 5%). Total macrophyte biomass in bioassays varied from 0.68 ± 0.19 to 0.84 ± 0.25 g in week 5 and from 1.42 ± 0.35 to 1.98 ± 0.02 g in week 22.

Table Table 7.. Results of no observed effect concentration (NOEC) calculations of the physicochemical parameters (drift percentage; Williams test, p < 0.05)a
DaypH decreaseOxygen surface decreaseDaypH increaseOxygen surface increase
  1. a Only endpoints showing a consistent response (NOECs calculated for 2 consecutive sampling dates) are displayed. Em dashes indicate NOEC ≥ 5%.

  2. b NM = not measured.

-4NMb88
10.2%91
20.2%92
30.2%94
41%98
90.2%99
141%1%101
150.2%105
161%1%106
21108
231121%
291131%
36115
421191%
43120
491%1231%
501%126
531%127
56130
57133
60134
63137
64140
67141
701%144
71147
74148
77151
78154
81161
84169
85176

Leaf litter decomposition

Significant treatment-related effects on leaf litter decomposition were not observed in the 7 decomposition experiments we performed (NOECs > 5%). Only at week 8 was a single NOEC of 1% calculated. The decay rate varied from 19.29 ± 2.24% in 5% treatments to 21.78 ± 1.75% in 0.2% treatments over a 2-week period. In the control ditches, the decay rate was 20.63 ± 2.46% over a 2-week period.

Figure Figure 4..

Principal response curves with species weights (bk) for the macroinvertebrate data set, indicating the effects of applications of compounds of the pesticide package. Of the variance, 38% could be attributed to sampling date and is displayed on the horizontal axis. Differences between replicates accounted for 38% of the variance. Twenty-four percent of the variance could be attributed to the treatment scenario. Of this percentage, 34% is displayed on the vertical axis. The species weight can be interpreted as the affinity of the taxon to the principal response curve. Taxa that have a species weight between −0.5 and 0.5 have a low correlation with the response curve and are therefore not displayed. The PRC diagram does display a significant amount of the treatment variance (p = 0.003). Application of herbicides, medium shading; application of insecticide, light shading; application of fungicides, solid.

DISCUSSION

Experimental design

The treatment regime of the ditch mesocosms was considered a worst case, despite the absence of herbicide application at the end of the season. Instead of this, mechanical removal was simulated because of limitations set by the ditch mesocosms (no persistent compounds could be applied) and limitations set by the compounds (compounds should be permitted for use in The Netherlands at that moment and in near future). Because ditch mesocosms were not connected to each other during the experiment, the absence of recovery via water contributed to the worst-case situation. Recovery from neighboring ditches via air was, of course, possible and reflected more or less the field situation.

Exposure concentrations

Except for prosulfocarb, the dissipation of other pesticides from the water of the ditch mesocosms was very rapid, and concentrations generally reached detection limits before the next application was made (Figure 2 and Table 5). The field dissipation measured covers the combined aqueous losses owing to photolysis, hydrolysis, sorption, volatilization, and biodegradation. Dissipation DT50 values from the literature (where available, field values) were compared to those measured in this experiment. For prosulfocarb, no other field studies were available and in the standard laboratory tests DT50 is 1 d in water and 150 to 380 d in sediment in a sediment-water study (data from Syngenta). For metribuzin, a dissipation half-life of 5 d was found in experimental pond mesocosms by Fairchild and Sappington (2002), whereas a range of 6 to 9.4 d (mean, 7.1 d) was reported by Brock et al. (2004) in a ditch enclosure study. In the present study, the dissipation of metribuzin from the water layer was faster than reported elsewhere. One possible explanation for this could be differences in photolysis, because this process is reported to be important for the dissipation of metribuzin (Muszkat et al. 1998, 2002).

For lambda-cyhalothrin, DT50 values ranged from 0.7 to 1.1 d in laboratory microcosms (Van Wijngaarden et al. 2004) and from 0.24 to 0.27 d in ditch enclosures (Leistra et al. 2003; Roessink et al. 2005). Wendt-Rasch et al. (2004) calculated a mean DT50 in water of 0.86 d in outdoor Elodea-dominated microcosms and a mean water DT50 of 2.4 d in outdoor Lemna-dominated microcosms. Hand et al. (2001) found a DT50 of less than 0.125 d for dissipation from the water column in an indoor macrophyte-dominated microcosm study and a DT50 of less than 0.125 d for the whole system. The dissipation of lambda-cyhalothrin in the present study was in the same range as reported by Van Wijngaarden et al. (2004) and in the Elodea-dominated microcosms reported by Wendt-Rasch et al. (2004) but was slower than that reported by Leistra et al. (2003), Roessink et al. (2005), and Hand et al. (2001). Alkaline hydrolysis in the water near the surface of macrophytes is considered to be the main dissipation process for lambda-cyhalothrin (Leistra et al. 2003).

Available water DT50 values for fluazinam range from 1.5 to 3.3 d in indoor microcosms (Van Wijngaarden et al. 2004) and from 0.9 to 1.1 d in outdoor microcosms (Wendt-Rasch et al. 2004). The dissipation of fluazinam in our study fell within these ranges. For chlorothalonil, field DT50 values in water range from 0.17 (Davies 1988) to 1.25 d (Ernst et al. 1991). In the present study, the dissipation of chlorothalonil was somewhat slower than that reported by Davies (1988).

Table Table 8.. Results of the Monte Carlo permutation test (p value) and no observed effect concentrations (NOECs) on the zooplankton, phytoplankton, and macroinvertebrate community level (drift percentage; Williams test, p < 0.05) for the different treatment levels of a pesticide package containing prosulfocarb, metribuzin, lambda-cyhalothrin, chlorothalonil, and fluazinam
 ZooplanktonPhytoplanktonMacroinvertebrates
WeekpNOECpNOECpNOEC
−1>0.10>5%>0.10>5%>0.10>5%
2>0.10>5%0.08>5%
3>0.10>5%
50.080>5%0.041%
7≤0.0010.2%
80.0091%>0.10>5%
110.0201%>0.10>5%0.0051%
140.0151%>0.10>5%
150.0171%
17>0.10>5%>0.10>5%
190.0081%
20>0.10>5%>0.10>5%
230.0011%
24>0.10>5%>0.10>5%

In conclusion, when compared with the field DT50 values reported in other studies, the fate of the compounds was overall very comparable. The relatively fast dissipation of metribuzin, lambda-cyhalothrin, chlorothalonil, and fluazinam from the water of the ditch mesocosms suggests that in our test systems these pesticides cause short-term stress only, despite the repeated application of some of these compounds. However, the longer exposures to the herbicide prosulfocarb may have also led to chronic stress on the phytoplankton community.

The 2nd application of lambda-cyhalothrin, which was approximately 50% lower than intended, had no consequences for the 5% treatment. Concentrations still were above threshold levels (Tables 1 and 11). It might have had consequences for the 1% treatment, because realized nominal concentrations were around threshold concentrations. If the target concentration (just above threshold concentration) would have been realized instead, the effect at the 1% treatment level might have been consistent between weeks 5 and 9 (community-level NOECmacroinvertebrates at the 0.2% treatment level [Table 8] on both dates and the effects would have been short lived). However, this would not have changed the results and conclusions of this experiment: a NOEC for the community of 0.2% and clear but transient effects at the 1% treatment level. Effects of lambda-cyhalothrin are in line with effects found in other mesocosm experiments (Van Wijngaarden et al. 2004; Roessink et al. 2005).

Secondary effects and interaction

In addition to direct effects on sensitive species, pollutants may exert effects on tolerant species by a number of ecological mechanisms. Such effects are called indirect (or secondary) contaminant effects (Fleeger et al. 2003). Competitive release and trophic cascades are indirect effects that are very common (Fleeger et al. 2003). Only in studies at the population, community, or ecosystem level can indirect contaminant effects be detected. Because indirect effects may be an important part of contaminant effects at these levels, we will consider some important indirect effects that were detected in the experiment more profoundly.

The trend of an effect on phytoplankton after the application of prosulfocarb in week 2 (Table 8), the observed effect on phytoplankton in week 5 (Table 8), and effects on Flagellatae (increase) in weeks 2 and 5 suggest that the effects observed can be attributed to herbicide application and not to the application of lambda-cyhalothrin. It seems most likely that these effects were a result of exposure to prosulfocarb (Table 12). The ecological reason for the increase in Flagellatae is difficult to explain from the data derived in the experiment. An increase in numbers usually indicates the occurrence of an indirect effect, such as a result of a decrease in other taxa. This primary effect, however, could not be demonstrated by the data. Chlorophyll-a concentrations of phytoplankton did not show a consistent treatment-related response, despite the temporal increase in Flagellatae. The decreases in pH and dissolved oxygen described above suggest that prosulfocarb decreases the net photosynthesis of the community of primary producers at the 5% treatment level in particular. However, consistent declines in populations of primary producers could not be observed during the first 5 weeks after prosulfocarb application. In addition, significant declines in zooplankton and macroinvertebrate taxa were also not observed in this period. A decrease in pH was also observed in the 8th week of the experiment (Table 7). A decrease in pH suggests a decrease in the net photosynthesis of the primary producers. Increases in oxygen levels, turbidity, and chlorophyll-a in water in weeks 14 to 16 (Table 7) suggest an increased algal production. However, effects on algae community structure were not observed in the 3 examples mentioned above. One possibility might be that effects on functional endpoints last longer and can be detected, whereas effects on structural endpoints might be too transient to detect with the chosen sampling frequency. A 2nd possibility might be the filtration of the samples through a net with 40-μm mesh size. As a consequence of this, the smallest species (among others, nanoplankton) are not considered in this experiment. Therefore, effects on phytoplankton in this experiment are limited to the larger species. The decrease in filamentous algae from week 13 onward and the persistence of this effect at the 5% treatment level may be another secondary effect, of which the cause is not clear.

Figure Figure 5..

Dynamics of macroinvertebrate populations showing consistent treatment-related responses (Table 9) after applications of compounds of the pesticide package. Numbers per artificial substrata are geometric mean abundance numbers of Gammarus pulex (A), Chaoborus sp. (B), Caenis horaria (C), Cloeon dipterum (D), Haliplidae (E), Dero sp. (F), Lymnaea stagnalis (G), Orthocladiinae/Chironomini (H), and Stylaria lacustris (I).

Some Rotifera showed clear direct effects (Figure 6). The decrease in the population of P. remata may be owing to direct effects of chlorothalonil, but indirect effects may also play a role. Because of the observed indirect effect of increasing daphnid abundance owing to a decrease in its predator Chaoborus, competitive exclusion may have led to a decline in Rotifera populations.

The significant reductions in G. pulex at the 5% treatment level did not influence the extent to which leaf litter was decomposed in the ditch mesocosms consistently. Only in week 8 was a single NOEC of 1% calculated. A reduced decomposition can probably be attributed to the effect of lambda-cyhalothrin, but the effect was not consistent. Obviously, other shredders such as Proasellus meridianus/coxalis, which were unaffected by the pesticide package, were able to maintain this function.

The adverse effect on Chaoborus obscuripes (Figures 4 and 5B) contributed to several indirect effects. In the 0.2% and 1% treatments, C. obscuripes was eliminated. Hence, predation of C. obscuripes did not reduce the numbers of Calanoida at these treatment levels (Figure 7D). At the 5% level, Calanoida show a strong decrease owing to direct effects after treatment with lambda-cyhalothrin and chlorothalonil. In the controls, numbers of Calanoida were reduced by predation of C. obscuripes. It may be owing to such nonlinearity in the observed responses that apparent effects were absent.

Table Table 9.. Results of no observed effect concentration (NOEC) calculations of zooplankton, phytoplankton, macroinvertebrate taxa, and percentage cover of floating filamentous algae (drift percentage; Williams test, p < 0.05) for the different treatment levels of a pesticide package containing prosulfocarb, metribuzin, lambda-cyhalothrin, chlorothalonil, and fluazinama
 ZooplanktonPhytoplanktonMacrophytes
WeekHexarthra sp.Polyarthra remataDaphnia group galeataFlagellatae sp. (˜5–6 micron, 2 flag.)Filamentous algae
-1
21% ↑
51% ↑
81% ↓0.2% ↑
111% ↓1% ↑
141% ↓1% ↓
170.2% ↓
190.2% ↓
201% ↓
241% ↓
Macroinvertebrates
WeekGammarus pulexChaoborus sp.Caenis horariaCloeon depterumHaliplidaeDero sp.Lymnaea stagnalisOrthocladiinae/ChironominiStylaria lacustris
  1. a Only taxa showing a consistent response (NOECs calculated for 2 consecutive sampling dates) are displayed. Blank cells indicate NOEC ≥ 5%. ↓ = populations were significantly reduced at concentrations above NOEC; ↑ = populations were significantly increased at concentrations above NOEC.

-10.2% ↑
3
71% ↓0.2% ↓1% ↓1% ↑0.2% ↑0.2% ↑
111% ↓0.2% ↓1% ↓1% ↓1% ↑1% ↑
151% ↓0.2% ↓1% ↓1% ↑1% ↑1% ↑
191% ↓0.2% ↓1% ↑1% ↑1% ↑
231% ↓1% ↓1% ↑

Significant reductions in the abundance of the mayfly C. dipterum by lambda-cyhalothrin also resulted in several indirect effects. As a result of competitive release, other water column inhabitants like S. lacustris, Dero sp., and Orthocladiinae/Chironomini may have increased.

Other indirect effects observed in the ditch mesocosms were an increase of the flatworms Dugesia lugubris and Mesostoma and an increase of the snail Lymnaea stagnalis. The latter phenomenon has been observed in other mesocosm experiments (Van Wijngaarden et al. 2004; Roessink et al. 2005) and is probably owing to competitive release and subsequent higher food availability. Farmer et al. (1995) found higher algal biomass and higher abundances of copepod nauplii and Rotifera as indirect effects of lambda-cyhalothrin at levels above the highest concentration in our study. Effects were probably owing to effects on Asellidae and Gammaridae. In the present study, Asellidae did not decrease significantly.

Summary of effects and ecological threshold concentrations of individual compounds

In the present study, the total pesticide package resulted in short-term responses of pH, oxygen, and phytoplankton and in long-term effects on zooplankton and macroinvertebrates at the 5% treatment level. At the 1% treatment level, only slight, transient effects were observed on a limited number of zooplankton and macroinvertebrate species and on pH. At the 0.2% treatment level, no consistent treatment-related effects were observed.

The effects at the community level were compared with known ecological threshold concentrations (NOEC and Class 2 LOEC values) for the individual compounds based on microcosm and mesocosm studies with these individual compounds (Table 11 with data from Brock et al. 2004; Van Wijngaarden et al. 2004; Roessink et al. 2005; data not shown). This table shows that the effects of metribuzin and fluazinam at the community level are unlikely at the concentrations applied in the experiment. The treatment concentrations in the experiment lie below or in the range of the threshold concentrations for these compounds (for metribuzin, see Fairchild and Sappington 2002; Brock et al. 2004). Effects of chlorothalonil might be expected at the 5% treatment level but are unlikely at the 1% treatment level. The treatment concentrations of lambda-cyhalothrin lie above the thresholds, indicating that clear effects would be expected at both the 1% and 5% treatment levels. For prosulfocarb only the First Tier Uniform Principles Standard is available (8.6 μg/L), indicating that probably effects would be expected at both the 1% and 5% treatment levels.

Figure Figure 6..

Principal response curve (PRC) with species weights (bk) for the zooplankton data set, indicating the effects of applications of compounds of the pesticide package. Of the variance, 44% could be attributed to sampling date and is displayed on the horizontal axis. Differences between replicates accounted for 38% of all variance. Eighteen percent of the variance could be attributed to treatment scenario. Of this percentage, 24% is displayed on the vertical axis. The species weight can be interpreted as the affinity of the taxon to the PRC. The PRC diagram does display a moderate significant amount of the treatment variance (p = 0.068). Application of herbicides, medium shading; application of insecticide, light shading; application of fungicides, solid.

Table 12 provides an overview of the effects of the treatments on the various endpoints studied. In this table, observed effects were summarized into effect classes described by Brock, Lahr, et al. (2000) and Brock, Van Wijngaarden, et al. (2000). At the 5% level, a strong decline was observed for G. pulex, Chaoborus, and mayfly larvae, most likely caused by direct toxic effects of lambda-cyhalothrin (Tables 1 and 11). Also Haliplidae decreased at the end of the experiment. Increase in numbers was observed in populations of Dero sp., L. stagnalis, Orthocladiinae/Chironomini, and S. lacustris. This may be, at least partly, a result of competitive exclusion due to the strong negative effect on the mayfly C. dipterum by lambda-cyhalothrin. During the period of chlorothalonil application, decreases in population densities of Rotifera and Calanoida were observed. At the 5% treatment level, the ecological threshold concentrations for chlorothalonil were exceeded (Tables 1 and 11). The effects on Rotifera and Calanoida can therefore most probably be attributed to exposure to chlorothalonil. Although based on laboratory data, effects of fluazinam on the zooplankton might have been expected. However, they were not observed in this experiment. At the community level, effects of fluazinam are unlikely (Tables 1 and 11). Effects on the higher levels might be absent because of functional redundancy between species. At the 5% level, a short-term decrease in pH and DO and an increase of Flagellata were observed in the 1st weeks of the experiment. Because effects of metribuzin at the community level are unlikely at the concentrations applied in the experiment (Tables 1 and 11), these effects can be attributed to prosulfocarb. Filamentous algae were lower than were the controls from week 13 onward. In weeks 14 to 15, algal productivity was increased. So, at the 5% treatment level, pronounced effects were observed and could be attributed to 3 of the 5 compounds within the pesticide package. Most of these effects persisted until the end of the experiment (effect class 4 in Table 12). Recovery was observed within the group of insects (Figure 5B to D), which may be attributed to the multivoltine character of their propagation and their recolonization abilities by air from neighboring control ditches. Recolonization by other routes was not possible in this experiment because of the hydrological isolation of the ditches. Hence, recovery of the population of G. pulex could not occur because of a lack of external supply. Other macroinvertebrates without a terrestrial stage in their life cycle were also hindered to recover. Rotifera and Calanoida recovered rapidly as a result of survival in less-affected microhabitats or production of resistant eggs. The importance of life cycle characteristics to the recovery potential of affected habitats and ditches has also been stressed by other authors (Maund et al. 1997; Sherratt et al. 1999).

Figure Figure 7..

Dynamics of zooplankton populations showing consistent treatment-related responses (except for Calanoida, Table 9) after applications of a pesticide package containing prosulfocarb, metribuzin, lambda-cyhalothrin, chlorothalonil, and fluazinam. Numbers are geometric mean abundance numbers of Hexarthra sp. (A), Polyarthra remata (B), Daphnia group galeata (C), and Calanoida (D).

At the 1% treatment level, transient decreases were observed in populations of Chaoborus and C. horaria, most likely caused by lambda-cyhalothrin (Tables 1 and 11), and on Hexarthra sp., likely caused by lambda-cyhalothrin and chlorothalonil (Table 9). A clear negative effect was observed on Haliplidae at the end of the experiment. Clear increases of population numbers of L. stagnalis and Orthocladiinae/Chironomini were observed, although abundance values of these invertebrates were small (Figure 5G and H). Another clear indirect effect is the temporary increase of Cladocera, most likely caused by the direct toxic effects of lambda-cyhalothrin on the predator Chaoborus. Prosulfocarb most probably caused a short-term decrease in pH in the 1st weeks of the experiment. Floating filamentous algae decreased from week 13 onward but were not significantly different any more from controls and 1% treatments by the end of the experiment. In weeks 14 to 15, algal production is increased. We can conclude that at the 1% level only slight, short-term effects were observed.

For macroinvertebrates a NOEC was calculated at the treatment level of 0.2%. The NOECs for other invertebrates (zooplankton) and phytoplankton were equal to a treatment level of 1%. For the pesticide package, the most sensitive organisms within the community were the macroinvertebrates, with Chaoborus and C. horaria as the most sensitive species and lambda-cyhalothrin as the most toxic compound. Hence, overall, we consider the 0.2% treatment level as the NOEC for the community.

Table Table 10.. Dynamics of submerged and floating filamentous algae expressed as mean cover percentage per treatment level per sampling date plus associated standard deviation
Submerged filamentous algaeControl0.2%1%5%
  1. a Asterisks indicate significant differences (Williams test, p < 0.05).

-129.47 ± 25.546.94 ± 7.5120.83 ± 36.086.67 ± 3.57
721.91 ± 25.424.19 ± 5.921.06 ± 1.062.31 ± 2.31
1321.78 ± 19.5021.38 ± 19.279.58 ± 16.0631.00 ± 24.00
1914.50 ± 7.1219.81 ± 4.684.38 ± 5.568.25 ± 7.72
2326.75 ± 13.7420.31 ± 1.334.88 ± 4.28*3.13 ± 1.44*a
Floating filamentous algae
-18.44 ± 7.672.38 ± 3.0110.42 ± 16.870.83 ± 1.44
75.97 ± 4.376.04 ± 4.719.56 ± 9.562.19 ± 2.19
1315.06 ± 11.0114.25 ± 4.958.21 ± 14.004.94 ± 1.81
1919.88 ± 6.8421.06 ± 4.336.79 ± 6.37*5.54 ± 0.71*
2314.41 ± 3.6523.06 ± 6.288.46 ± 9.902.54 ± 1.50*
Table Table 11.. Threshold levels (no observed effect concentration [NOEC]) and lowest observed effect concentration causing slight effects (Class 2 LOEC) for several endpoints in microcosm and mesocosm studies with individual compoundsa
 Ecological threshold concentrations (NOEC - Class 2 LOEC)
EndpointsMetribuzin (μg/L)Lambda-cyhalothrin (ng/L)Chlorothalonil (μg/L)Fluazinam (μg/L)
  1. a NOEC = no observed effect concentration; Class 2 effects = slight effects; LOEC = lowest observed effects concentrations causing Class 2 effects; ↓ = populations were significantly reduced at concentrations above the thresholds; ↑ = populations were significantly increased at concentrations above the thresholds; ↓↑ = populations were both significantly reduced and significantly increased at concentrations above the thresholds.

  2. Metribuzin: Brock et al. (2004); lambda-cyhalothrin: Van Wijngaarden et al. (2004); Roessink et al. (2005) (macrophyte-dominated mesocosms); chlorothalonil: Syngenta, United Kingdom, unpublished data; fluazinam: Alterra, Wageningen, The Netherlands, unpublished data.

  3. b Asterisks indicate values below nominal concentrations used in this study.

Microcrustaceans18–56 ↓10–25 ↓*b3–10 ↓*2–10 ↓
Rotifers18–56 ↓↑10 ↓↑*10 ↓*2–10 ↓
Phytoplankton5.6–18 ↓>25010 ↓↑*10–50 ↑
Macrophytes>180>250100>250
Macrocrustaceans10–25 ↓*>30010–50 ↓
Insects10 ↓*100–300
Other macroinvertebrates100–250 ↓↑10050–250
pH, DO5.6–18 ↓>250100>50

Toxicity of the pesticide package

Effects on primary producers are likely at TUalgae > 0.1 (Brock, Lahr, et al. 2000). At the 5% level, a temporal exposure peak up to 0.9 TUalgae was observed (Figure 3), which resulted in short-term responses of pH, oxygen, and phytoplankton (Table 7). For algae, effects of treatment predominantly resulted from prosulfocarb treatments and resulted in short-term responses of pH, oxygen, and phytoplankton.

After repeated application, effects on invertebrates are likely to occur at TUDaphnia < 0.1 (Brock, Van Wijngaarden, et al. 2000). At the 5% level, exposure concentrations amounted to 0.4 TUDaphnia and resulted in long-term responses (see Figures 4 and 6). For invertebrates, effects were probably mainly caused by lambda-cyhalothrin and chlorothalonil (Table 11). Effects on zooplankton were most apparent from week 8 onward, after the application of lambda-cyhalothrin in week 5 and applications of chlorothalonil in weeks 6 and 7 (Figures 3 and 6). The effects on zooplankton continued into the period in which fluazinam was sprayed. However, some zooplankton populations were already showing recovery within this period (Figure 7B and C). The calculations of TU show that treatment concentrations for fluazinam lie in the range in which effects are border line. This is in accordance with Tables 1 and 11, which show that effects of fluazinam are unlikely.

Table Table 12.. Summary of effects observed in ditch mesocosms treated with pesticides used in potato cultivationa
 Treatment level
Endpoint0.2%1%5%
  1. a Treatment levels are equal to 0.2%, 1%, and 5% spray drift emission of label-recommended rates. The numbers in the table refer to effect classes described by Brock, Lahr et al. (2000) and Brock, Van Wijngaarden et al. (2000) and included in the European Union Guidance Document on Ecotoxicology (SANCO 2002). PRC = principal response curve; 1 = no effect, 2 = slight effect, 3 = clear short-term effect, full recovery observed, 4 = clear effect, no full recovery observed at the end of the experiment, 5 = clear long-term effects; ↓ = decrease of endpoint; ↑ = increase of endpoint; ↓↑ = both decrease and increase of endpoint is observed.

Microcrustaceans12↑2↑
Rotifers12↓3↓
PRC Zooplankton12↓3↓
Phytoplankton (Flagellata)12↑3↑
Macrophytes (Filamentous algae)13↓4↓
Macrocrustaceans114↓
Insects12↓3↓
Other macroinvertebrates12↓↑4↓
PRC macroinvertebrates12↓4↓
pH/dissolved oxygen12↓3↓
Figure Figure 8..

Dynamics of Flagellatae sp. population belonging to the phytoplankton. This taxon is the only taxon that showed a consistent treatment-related response (Table 9).

Effects based on multi- and repeated stress

The exposure concentrations in toxic units (Figure 3) may indicate potential risks owing to multi or repeated stress. During the experiment 2 periods of multi-stress can be recognized. The first is the application period of the 2 herbicides, in which multistress of the 2 herbicides seems to exhibit a potential risk. However, it is not an actual risk, because effects of metribuzin at the community level are unlikely at the concentrations applied in the experiment (Tables 1 and 11) and effects can be attributed to prosulfocarb. The 2nd is the simultaneous application of lambda-cyhalothrin and chlorothalonil as indicated by Figure 3. Experimental data seem to confirm an effect by the 2 stressors. The maximum TU was reached after week 9 (Figure 3) after the application of both lambda-cyhalothrin and chlorothalonil. This was followed by a clear, but not significant, effect on zooplankton at the 1% level (Figure 6). This effect was not observed after the application of 1 of the 2 compounds separately in the weeks before. Similar effects at the 1% level are also clear at the population level within the population of Hexarthra sp. (Figure 7 and Table 9). At the 5% level, the effects proceed until recovery starts, which was not until week 15 (Figure 6). It can be concluded that at the 1% level minor effects on zooplankton were observed, which might result from multistress by lambda-cyhalothrin and chlorothalonil. These effects on zooplankton and macro-invertebrates seem to be more severe at the 5% level.

During the experiment repeated stress was potentially caused by the multiple applications of lambda-cyhalothrin and chlorothalonil. Repeated stress might have played a role in the effects of chlorothalonil on zooplankton and in the effects of lambda-cyhalothrin on macroinvertebrates, although in the experiment these effects cannot be separated from the effects of the combined application of chlorothalonil with lambda-cyhalothrin on the 4th chlorothalonil application. We can conclude that multi- and repeated stress played a small role within the applied pesticide package. This was probably mainly because most of the substances rapidly dissipated and the absence of many simultaneous applications.

CONCLUSIONS

Comparing the findings from previous experiments with those observed in the present study, it can be concluded that the patterns of effects observed from the application of the pesticide program were in line with the effects we could deduce from the thresholds and concentrations for the individual compounds, although repeated applications were used. This may be because of the short dissipation time of most compounds.

If no effects on aquatic ecosystems are accepted and if spray drift is considered the only emission route, emission reduction measures to values below 1% spray drift are necessary. The current aquatic risk assessment procedure, based on individual compounds, the Uniform Principles, and a drift emission of 1%, may sufficiently protect aquatic ecosystems if slight and transient effects are accepted. The crop approach is a promising approach for risk assessment and evaluation of effects of realistic pesticide stress. Moreover, a crop approach is also promising from a risk management point of view, because it will improve the practicality and acceptance.

Acknowledgements

The present study was financially supported by the Ministry of Agriculture, Nature Conservation, and Food Safety and by Syngenta Crop Protection. We thank Steven Crum, Ariënne Matser, Alex Schroer, Jos Sinkeldam, Marie-Claire Boerwinkel, and Frans van Tilburg.

Ancillary