JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Júnior, R. P. S.
Right arrow Articles by van der Zee, S. E. A. T. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Júnior, R. P. S.
Right arrow Articles by van der Zee, S. E. A. T. M.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Júnior, R. P. S.
Right arrow Articles by van der Zee, S. E. A. T. M.
Related Collections
Right arrow Agricultural Pesticides
Right arrow Pesticides
Right arrow Nonequilibrium Flow
Right arrow Vadose Zone Processes and Chemical Transport
Published in J. Environ. Qual. 33:1473-1486 (2004).
© ASA, CSSA, SSSA
677 S. Segoe Rd., Madison, WI 53711 USA

TECHNICAL REPORTS

Vadose Zone Processes and Chemical Transport

Preferential Flow of Bromide, Bentazon, and Imidacloprid in a Dutch Clay Soil

Rômulo Penna Scorza Júniora, Johan H. Smeltb, Jos J. T. I. Boestenb,*, Rob F. A. Hendriksb and Sjoerd E. A. T. M. van der Zeec

a Embrapa Agropecuária Oeste, Caixa Postal 661, CEP 79804-970 Dourados, MS, Brazil
b Alterra, Wageningen University and Research Centre, PO Box 47, 6700 AA Wageningen, the Netherlands
c Department of Environmental Sciences, Wageningen University and Research Centre, PO Box 8005, 6700 EC Wageningen, the Netherlands

* Corresponding author (jos.boesten{at}wur.nl).

Received for publication May 9, 2003.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Leaching to ground water and tile drains are important parts of the environmental assessment of pesticides. The aims of the present study were to (i) assess the significance of preferential flow for pesticide leaching under realistic worst-case conditions for Dutch agriculture (soil profile with thick clay layer and high rainfall) and (ii) collect a high-quality data set that is suitable for testing pesticide leaching models. The movement of water, bromide, and the pesticides bentazon [3-isopropyl-1H-2, 1,3-benzothiadiazine-4(3H)-one-2,2-dioxide] and imidacloprid [1-[(6-chloro-3-pyridinyl)-methyl]-N-nitro-2-imidazolidinimine] was monitored in a clay soil for about 1 yr. The 1.2-ha field was located in the central part of the Netherlands (51°53' N, 5°43' E). The soil was a Eutric Fluvisol cropped with winter wheat (Triticum aestivum L.). Tile drains were present at a 0.8- to 0.9-m depth and the ground water level fluctuated between a 0.5- and 2-m depth. All chemicals were applied in spring. None of the soil concentration profiles showed bimodal concentration distributions. However, for each substance the highest concentration in drain water was found in the first drainage event after its application, which indicates preferential flow. This preferential flow is probably caused by permanent macropores that were present in the 0.3- to 1.0-m layer. At the time of the first drainage event, the drain water concentration of each substance was about an order of magnitude higher than its ground water concentration. Thus, the flux concentrations in drain water proved to be a more sensitive detector of preferential flow than the resident concentrations in the soil profile and the ground water.

Abbreviations: HPLC, high performance liquid chromatography • LOQ, limit of quantification


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
THERE IS A WORLDWIDE CONCERN about ground water contamination by pesticides used in agriculture. Different studies have shown that pesticides are present in ground water in the United States and Europe (Hallberg, 1989; Clark et al., 1991; Domagalski and Dubrovsky, 1992; Legrand et al., 1992). Leaching from the soil profile is the major cause for the occurrence of pesticides in ground water (Flury, 1996). First screening assessments of pesticide leaching to ground water have been developed assuming convective–dispersive transport (Jury et al., 1987). However, the presence of macropores (e.g., shrinkage cracks, earthworm, and root channels) in clay soils can enhance the leaching of surface-applied pesticides (Larsson et al., 1999) by creating high-conductivity flow pathways, which bypass most of the chemically and biologically reactive topsoil (Beven and Germann, 1982; Bouma, 1991). Also, for less structured soils there are many research results indicating that preferential flow plays an important role in the leaching of pesticides (Ghodrati and Jury, 1992; Flury, 1996). Preferential flow may also lead to relatively fast leaching of pesticides to surface water when tile or mole drains are present in the soil (Steenhuis et al., 1990; Kladivko et al., 1991; Harris et al., 1994; Larsson and Jarvis, 1999). Contamination of surface water is also an important point of concern in pesticide registration (FOCUS, 2001).

Scenario calculations with leaching models play an increasing role in the assessment of pesticide leaching to ground water and surface water for pesticide registration in the European Union (EU) and the USA. However, the validation status of these leaching models (including those describing preferential flow through macropores) still needs improvement; there is a wide range of soil, climatic, and agricultural conditions of the USA and the EU that is not fully covered by the available field tests of the models (Boesten, 2000a). An additional problem is that the EU drinking water limit of 0.1 µg dm–3 corresponds with leaching of only about 0.01 to 0.1% of a dose of 1 kg ha–1. This implies that field studies in which only 0.01 to 0.1% of the dose leaches are the most relevant for model tests with the appropriate sensitivity (Vanclooster et al., 2000). However, such data sets are difficult to generate. Another prerequisite is that a data set should consist of many different types of measurement. Armstrong et al. (1996) recommend a stepwise model testing and calibration procedure, that is, to first test and calibrate water flow and soil temperatures, next to test and calibrate tracer behavior, and only then start with testing of pesticide behavior. Thus, data sets should ideally contain (i) soil temperatures, (ii) moisture profiles, (iii) ground water levels, (iv) drain flow rates and percolation rates to ground water, (v) soil concentration profiles of a tracer, (vi) drain water concentrations of a tracer, (vii) ground water concentrations of a tracer, (viii) soil concentration profiles of a pesticide, (ix) drain water concentrations of a pesticide, (x) ground water concentrations of a pesticide, and (xi) laboratory studies on degradation rate and sorption with the soil and pesticide considered.

We analyzed five field data sets on pesticide leaching in clay soils (Harris et al., 1994; Brown et al., 1995; Johnson et al., 1996; Larsson and Jarvis, 1999; Jones et al., 2000) and checked which types of measurements described in the above list were available. The average score of positive answers was below 50% except for Larsson and Jarvis (1999), whose score was 80%. However, Larsson and Jarvis (1999) measured the leaching of a very mobile pesticide in winter. In their case, in the order of 10% of the dose leached, which is high compared with the 0.01- to 0.1%-level that is more relevant for testing of models for leaching to ground water with the sensitivity that complies with water quality standards. Thus, there appears to be scope for collecting high-quality data sets on pesticide leaching to ground water.

The objectives of the study were to (i) assess the significance of preferential flow for pesticide leaching under realistic worst-case conditions for Dutch agriculture and (ii) collect a high-quality data set suitable for testing pesticide leaching models. Thus, the leaching of two pesticides (bentazon and imidacloprid) was studied via a field experiment on a clay soil. Bentazon and imidacloprid show different behavior in soil. Bentazon is very mobile in soil and dissipates from field soil with a typical half-life of about 30 d (Boesten and van der Pas, 2000), whereas imidacloprid is moderately sorbed (Cox et al., 1997) and is very persistent in soil (Baskaran et al., 1999). The realistic worst-case conditions were achieved by selecting a cracking clay soil with a clay content in the range of 30 to 35% to a 1.2-m depth, where most Dutch clay soils consist of a thinner clay layer overlying a sandy layer. To ensure a high quality of the data set, the criteria set by Armstrong et al. (1996) were fulfilled as much as possible.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Experimental Field and Chemical Application
The field experiment was performed from April 1998 until April 1999 in the municipality of Andelst (51°53' N, 5°43' E, 8 m above sea level) in the Netherlands (Fig. 1). The soil is a young Holocene river bank deposit of the river Rhine classified as Eutric Fluvisol (Food and Agriculture Organization, 1998). Its clay layer was about 3 m deep and underlaid with a thick layer of coarse sand. After dry periods, shrinkage cracks were observed at the soil surface of the entire experimental field. Permanent macropores as a result of root decay and earthworm activity were regularly found in the 30- to 100-cm soil layer during the sampling of the soil profile. Soil profile characteristics are given in Table 1.



View larger version (22K):
[in this window]
[in a new window]
  		Fig. 1. (A) Map showing study site location in Andelst (the Netherlands) and (B) schematic diagram of the experimental field. In the diagram, the rectangles coded A to J are the 16 plots and the dashed lines indicate the six drain pipes.

 

View this table:
[in this window]
[in a new window]
  		Table 1. Physicochemical properties of the soil at the experimental field. Percentages are by mass.

 
The experimental study site was 160 m long and 50 m wide and was divided into 16 plots each 40 m long and 10 m wide (Fig. 1). Each plot was divided into 13 subplots with widths of 3 to 4 m. The slope of the field was 0.6% in the northeast direction. The experimental field had tile drains at a 0.8- to 0.9-m depth with lengths varying from 185 to 210 m and a spacing of 10 m. The experimental field comprised the entire catchment area of six tile drains. Three adjacent drain outlets were coupled, which resulted in two drain sets for measurements. In October 1997, four ground water sampling tubes with filter depths of 0.7 to 0.9, 1.0 to 1.2, 1.3 to 1.5, and 1.9 to 2.8 m were installed in the center of each plot 0.5 m apart in a row parallel to the drains. The polyethene tubes (5-cm i.d.) had filters consisting of a slotted plastic screen at the bottom (which was closed with a stopper). The tubes were installed via drilling holes (10-cm diameter) to the desired depths. Filter sand was applied around the filters and each bore hole was further filled up with bentonite clay.

On 23 Oct. 1997 winter wheat was sown. Bentazon and bromide were applied on 7 Apr. 1998 in two separate spraying events with a boom sprayer. The bentazon dose was 1.33 kg ha–1 (applied in 380 dm3 water ha–1 as Basagran P Duplo; BASF AG, Ludwigshafen, Germany). We applied KBr at a bromide dose of 59.6 kg ha–1 in 600 dm3 water ha–1. Within 12 h after bentazon and bromide application, 5.7 mm of rain fell. An imidacloprid dose of 0.7 kg ha–1 (as Admire; Bayer AG, Leverkusen, Germany) was applied in 600 dm3 water ha–1 on 27 May 1998. On this day cracks were visible in the soil surface. On 29 May 1998 about 15 mm of rain fell during the first 5 h of the day. The ground water sampling tubes and a circle of 0.5 m around the tubes were shielded by plastic foil during spraying to prevent direct contamination. The winter wheat was harvested on 20 Aug. 1998. The wheat residue was chopped and left at the soil surface, so no crop residues were removed from the field. On 8 Dec. 1998 the field was plowed (0.25–0.30 m depth) and again winter wheat was sown.

Monitoring of Meteorological Data, Temperatures, Ground Water Levels, and Drain Discharge
Rainfall was recorded with a gauge at the experimental field. The surface area of its aperture was 200 cm2 and its rim was 40 cm above the soil surface. It operated via a tipping bucket with a volume corresponding to about 0.2 mm of rain and was connected to a data logger, which recorded each tipping time. The measured rainfall was calculated back to rainfall reaching the soil surface using a second gauge whose rim was flush with the soil surface. Daily values of minimum and maximum air temperature, global radiation, average wind speed, and air humidity were obtained from the meteorological station of Wageningen University and Research Center located 10 km from the experimental field. Soil temperatures were measured using thermistors in duplicate at 5-, 50-, and 100-cm depths at the experimental field. Ground water levels were continuously measured at two locations in the experimental field using tubes with filters reaching a 235-cm depth and pressure transducers. The hydraulic head in the coarse sand layer was manually measured at approximately 2-wk intervals using tubes with 1-m-long filters between depths of 5 and 6 m. The tubes for measuring the hydraulic head were installed via drilling holes to the desired depths. Filter sand was applied around the filters and each bore hole was further filled up with bentonite clay. The discharge of the two drain sets was measured via measuring the pressure head in an open channel with a V-notch sharp-crested weir with an angle of 0.5 rad (Smelt et al., 2003).

Sampling of Soil Profiles, Drain Water, and Ground Water
Soil profiles were sampled at 1, 22, 52, 69, 125, 167, 239, and 378 d after bromide and bentazon application (imidacloprid was applied 50 d after bromide and bentazon application). On Days 1 and 52, only the top 15 cm was sampled at 32 spots distributed over the field with a split-tube auger (30-cm length and 9.5-cm i.d.). This was done to check the amounts of bromide, bentazon, and imidacloprid that had entered the soil after the first significant rainfall events considered as estimates of the amounts that reached the soil. The 15-cm soil cores were sliced into 5-cm layers. On all other sampling dates, soil was sampled up to a 120-cm depth from the 16 plots (one sample per plot) using a steel corer (9.6-cm i.d. and 120-cm long) that was built to collect soil cores in PVC tubes (used as liners inside the corer). This corer was pressed vertically into the soil with a hydraulic wheeled excavator. The 16 PVC tubes containing the soil were transported to the laboratory where they were kept at room temperature. Within 2 d, the soil cores were sliced in 10-cm sections starting from the bottom of the core. Each 10-cm core section was put in a vertical position. The center 7.5 cm of the 9.6-cm core was cut out using a cutting ring and was used for analyses of bromide and pesticides. This was done to prevent measurement of artificial downward movement along the wall of the PVC tube (resulting from pressing the tube in the soil). The remaining outer part of the soil column was used for determination of the moisture content by drying at 105°C. The discharge of each drain set was proportionally sampled using a cooled ISCO (Lincoln, NE) sampler (Model 3700R). The temperature range in the sampler was 4 to 10°C and samples were transferred to the laboratory at 1- or 2-wk intervals and there stored at –20°C. On all 16 plots ground water samples were collected from the upper ground water layer at various times. Sampling depth depended on the actual ground water level. Occasionally, additional ground water samples were collected from deeper filters at the same time. Filters were well flushed before 0.5 dm3 of water was sampled in a glass flask via applying suction. The flasks were transferred to the laboratory and stored at –20°C until analysis. Soil and water, which were sampled before spraying the chemicals and chemical analysis, showed negligible levels of bromide and no detectable residues of bentazon and imidacloprid.

Procedures for Chemical Analysis and Extraction
The 16 soil columns of each sampling time were not analyzed individually, but the corresponding layers of four plots (ABCD, EFGH, IJKL, and MNOP; Fig. 1) were mixed to reduce analytical efforts. Similarly, ground water samples from the same filter depths of four plots were mixed for pesticide analysis. Bromide ion was extracted from soil by shaking 100 g of moist soil with 100 or 150 cm3 of a CaCl2 solution (0.01 mol dm–3) for 1 h. The bromide concentration in the supernatant was measured with high performance liquid chromatography (HPLC) and UV detection at 210 nm (Smelt et al., 2003). The extraction efficiency ranged from 90 to 110% and the limit of quantification (LOQ) for bromide in soil was 0.3 mg kg–1. The bromide concentration in ground water and drain water was analyzed with the same HPLC method. The LOQ was 0.25 mg dm–3. Bentazon was extracted from soil by shaking 100 g of moist soil with 100 cm3 of CaCl2 solution (0.01 mol dm–3) for 1 h. Bentazon concentrations were measured using HPLC and UV detection at 224 nm (Smelt et al., 2003). The extraction efficiency ranged from 82 to 100%. The LOQ for bentazon in soil was 0.5 µg kg–1. Imidacloprid was extracted from soil by shaking 100 g of moist soil with 100 cm3 of a mixture of acetone and water (80:20 by volume) for 1 h. The acetone was evaporated from the extract and imidacloprid was partitioned into dichloromethane, which was evaporated. The concentration of imidacloprid was measured using HPLC (Ishii et al., 1994; Fernandez-Alba et al., 1996). The extraction efficiency was on average 104% and the LOQ was 0.6 µg kg–1. Bentazon and imidacloprid were extracted from ground water and drain water using a solid–liquid phase extraction. Water samples of 500 to 800 cm3 were passed through a column packed with Bakerbond octadecyl (J.T.Baker, Phillipsburg, NJ). The concentrations were measured via HPLC as described for the analysis of the soil extracts. The extraction efficiency was on average 100% for bentazon and 92% for imidacloprid. The LOQ was 0.03 µg dm–3 for bentazon and 0.05 µg dm–3 for imidacloprid.

Crop Sampling and Analysis for Bromide Ion
The wheat crop was sampled on 11 May and 5 Aug. 1998 to determine the bromide content. Plants were collected from 8 to 16 spots spread over the field. Samples from two to four spots were mixed to reduce the number of extractions to four per sampling. The samples were dried and well mixed. They were extracted using HCl (0.01 mol dm–3). Extracts were analyzed with HPLC as previously described for bromide analysis in soil extracts.

Degradation Rate Experiments in the Laboratory
The degradation rate of bentazon was measured in soil samples taken from the 0- to 30-, 40- to 70-, and 80- to 120-cm layers on 30 Mar. 1998. The sample from the 0- to 30-cm layer was collected by taking 40 soil cores (6-cm i.d.) distributed over the study site. Soil samples from the 40- to 70- and 80- to 120-cm layers were collected from 16 undisturbed soil columns (one from each plot). The samples per layer were mixed and stored at 5 to 10°C for about 6 wk. Portions of 100 g of moist soil were added to glass jars and 4 cm3 of distilled water containing bentazon was added to each jar and mixed through the soil. The initial content of bentazon in the soil from the 0- to 30-, 40- to 70-, and 80- to 120-cm layers was 1.1, 0.11, and 0.01 mg kg–1, respectively. The soil samples were incubated at 5, 15, and 25°C for the 0- to 30-cm layer, at 10 and 15°C for the 40- to 70-cm layer, and at 10°C for the 80- to 120-cm layer. Moisture contents during incubation were 0.21, 0.23, and 0.27 kg kg–1 for the 0- to 30-, 40- to 70-, and 80- to 120-cm soil layers, respectively. Additionally, bentazon was incubated in soil from the 80- to 120-cm layer at 10°C under water-saturated conditions. This was done because the soil in this layer was exposed to both saturated and nonsaturated conditions during the experimental period. The initial concentration of bentazon in the water was 0.04 mg dm–3. The jars with soil from all incubation experiments were capped with aluminum foil in which a hole of 2 mm was made to ensure aerobic conditions. At increasing time intervals, bentazon was extracted from soil in duplicate jars by adding 100 cm3 of CaCl2 solution to the jars and shaking for 1 h. Bentazon was analyzed using the same procedure as used for the soil profile samples.

The degradation rate of imidacloprid was measured in soil samples taken on 3 Feb. 1999 from a 10-m-wide plot along the north side of the experimental field, which had been treated in the same way as the experimental field but had not been sprayed with pesticides in 1998. The soil of the 0- to 30-cm layer was sampled at 30 spots. The soil of the 40- to 70- and 80- to 120-cm layers was sampled at 10 spots using soil cores with inner diameter of 6 cm. Soil cores from the same layer were mixed and stored at 10°C for about 1 wk. Portions of 100 g of moist soil were added to glass jars and 4 cm3 of distilled water containing imidacloprid was added to each jar and mixed through the soil. Imidacloprid was incubated at 5, 15, and 25°C for the 0- to 30-cm layer, at 15°C for the 40- to 70-cm layer, and at 10°C for the 80- to 120-cm layer. The initial content of imidacloprid was 0.83 mg kg–1 for the 0- to 30-cm layer and 0.03 mg kg–1 for the 40- to 70- and 80- to 120-cm layers. Moisture contents during incubation were 0.22, 0.24, and 0.28 kg kg–1 for the 0- to 30-, 40- to 70-, and 80- to 120-cm layers, respectively. The jars were closed as described for bentazon. At increasing time intervals, imidacloprid was extracted from duplicate jars by adding 100 cm3 of a mixture of acetone and water (80:20 by volume). The extracts were analyzed as described for imidacloprid samples from the soil profiles.

Sorption Experiments
Sorption of bentazon was measured using the soil sample of the 0- to 30-cm layer from the degradation study. This sample had a pH (KCl) of 7.2, an organic carbon content of 1.28%, and a moisture content of 0.16 kg kg–1. Solutions of bentazon with concentrations of 0.1, 0.6, 3, and 14 mg dm–3 were made using CaCl2 solution (0.01 mol dm–3). A mass of 50 g of moist soil and 40 cm3 of a bentazon solution were added to centrifuge tubes (in quadruplicate) and these were rotated at a frequency of 0.3 s–1 for 24 h at 20°C. Thereafter, the tubes were centrifuged for 20 min at 20°C at 660 x g. The concentration of bentazon in the supernatant was measured using HPLC as described before. The sorption of imidacloprid was determined at 5, 15, and 25°C using the soil sample of the 0- to 30-cm layer from the degradation study. This sample had a pH (KCl) of 7.1, an organic carbon content of 1.45%, and a moisture content of 0.16 kg kg–1. Solutions of imidacloprid with concentrations of 0.1, 1, and 11 mg dm–3 were made in CaCl2 solution (0.01 mol dm–3). A mass of 50 g of moist soil and 50 cm3 of an imidacloprid solution was added to centrifuge tubes (in triplicate) and these were rotated at a frequency of 0.3 s–1 for 24 h at 5, 15, or 25°C. Thereafter, the tubes were centrifuged for 20 min at 660 x g at the same temperature as the sorption equilibration. The concentration of imidacloprid in the supernatant was measured using HPLC as described before. It was assumed that the decrease of the concentration in the liquid phase was completely attributable to sorption of bentazon or imidacloprid.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Availability of Data
Smelt et al. (2003) and Scorza Júnior (2002) give a more detailed description of the results. Smelt et al. (2003) give all results of the different laboratory experiments and field measurements in digital form (some 40 documented ASCII files). These data include original digital records of rainfall and drainflow intensities for the whole experimental period and standard deviations of all moisture and solute profiles, and they are freely available for scientific purposes from the corresponding author of this paper. Availability of relevant data in digital form has shown to be an important aspect of data sets in view of the large possible effect of modeller subjectivity on estimated pesticide-soil input parameters for models (Boesten, 2000b).

Laboratory Studies
Figure 2 shows that all degradation rate measurements (except those in Fig. 2B) could be described well with first-order kinetics. The corresponding half-lives (DT50) assuming first-order kinetics and using linear regression are given in Table 2. Figures 2A and 2B show that the temperature had a distinct effect on the degradation rates of bentazon and imidacloprid in soil from the 0- to 30-cm layer. Figure 2B shows that the degradation rate of imidacloprid slowed down after 100 d in the studies at 15 and 25°C. This can be possibly attributed to a decrease in the microbial activity in the incubation systems (Anderson, 1987). Therefore, imidacloprid data after 100 d in this graph were not used in the estimation of DT50. Figures 2A and 2B show that bentazon was transformed much faster in top soil layer than imidacloprid. The same holds for the two deeper layers. The effect of temperature on degradation of bentazon in the 40- to 70-cm layer is remarkable (Fig. 2C); the half-life at 10°C was more than a factor 2 longer than at 15°C (Table 2). For both pesticides the degradation rate decreased considerably with increasing depth (Fig. 2 and Table 2), which is probably related to a decrease in microbial activity with depth. Bentazon was only very slowly transformed in the 80- to 120-cm layer and the rate in the water-saturated soil was about equal to that in unsaturated soil (Fig. 2E and Table 2). From the degradation rate coefficients for the 0- to 30-cm soil layer, Arrhenius activation energies were calculated using linear regression. For bentazon, the energy was 74 ± 1 kJ mol–1 and for imidacloprid it was 64 ± 12 kJ mol–1. These energies are in the range found by FOCUS (1997) after reviewing 50 experiments with different soils and pesticides.



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 2. Remaining mass of bentazon and imidacloprid as a function of time as measured in laboratory incubation studies at different temperatures using soil from the experimental field. (A) Bentazon in the 0- to 30-cm layer, (B) imidacloprid in the 0- to 30-cm layer, (C) bentazon in the 40- to 70-cm layer, (D) imidacloprid in the 40- to 70-cm layer, (E) bentazon at 10°C in the 80- to 120-cm layer, and (F) imidacloprid at 10°C in the 80- to 120-cm layer. The points are the measurements and the lines are fits assuming first-order kinetics. In the calculation of the fitted lines for 15 and 25°C in (B), all measurements later than 100 d were ignored.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Half-lives (DT50) for bentazon and imidacloprid as measured in the laboratory at different temperatures using soil from the experimental field.

 
The sorption experiment with bentazon showed no significant change in the bentazon concentration; the concentration in the supernatant was on average 104 ± 7% of the initial value. This is consistent with the recovery of close to 100% of the extraction of soil with aqueous 0.01 mol dm–3 CaCl2 solution previously described (such an extraction can be considered as a desorption study). As the pKa of bentazon is 3.3 (Huber and Otto, 1994) and the pH (KCl) of our soil was 7.1 to 7.5 (Table 1), bentazon was present as an anion in soil solution. In clay soils, anion exclusion may be significant (Bolt and de Haan, 1979), which may even lead to "negative sorption." Weak or negligible adsorption of bentazon has also been reported by Abernathy and Wax (1973) and Boesten and van der Pas (2000). The imidacloprid sorption at different temperatures could be described well with the Freundlich equation. Fitted Freundlich coefficients and their standard deviations were 1.75 ± 0.03 dm3 kg–1 at 5°C, 1.51 ± 0.05 dm3 kg–1 at 15°C, and 1.21 ± 0.03 dm3 kg–1 at 25°C. These values refer to a reference concentration of 1 mg dm–3 (Boesten, 1994). The Freundlich exponents at the different temperatures were almost equal: 0.81 ± 0.008 at 5°C, 0.82 ± 0.014 at 15°C, and 0.81 ± 0.012 at 25°C. The measured Freundlich coefficients correspond with KOC values ranging from 83 to 121 dm3 kg–1. Cox et al. (1997) found KOC values for imidacloprid of 323 to 378 dm3 kg–1, Oi (1999) found values of 249 to 268 dm3 kg–1, and Capri et al. (2001) found values of 107 to 189 dm3 kg–1. So the KOC values of our soil are low in the range of values reported in the literature. Such a variability in KOC values is quite normal (Wauchope et al., 2002).

Water Flow in Soil
The ground water level varied from about 40 to 190 cm below the soil surface (Fig. 3A). The deepest level was reached around Day 140 (late August), which is a few days after harvest of the crop (on Day 135). The level rose sharply between about Days 150 and 160 when about 200 mm of rain fell (Fig. 3B). Thereafter the level fluctuated mostly between a 70- and 80-cm depth. The water level in the tubes that measured the hydraulic head in the coarse sand layer (data not shown in Fig. 3B) was usually deeper than or equal to the ground water level. However, on Days 209 and 210 this water level was 20 to 30 cm higher, thus indicating upward seepage from the subsoil. Cumulative rainfall during the entire experimental period was about 1150 mm (Fig. 3B). There was almost no drain discharge until about Day 150 (Fig. 3B). This is consistent with the ground water levels being usually below the drain depths (80–90 cm) in the first 150 d as shown in Fig. 3A. During spring and summer only one minor drainage event occurred; about 0.4 mm of water was measured in Drain Set 2 after heavy rainfall that occurred between Days 19 and 21. After Day 150 the two drain sets behaved similarly. However, the total discharge of Drain Set 2 was considerably larger than that of Drain Set 1 (466 versus 340 mm). Drain Set 2 always started to discharge water earlier than Drain Set 1 and, in most cases, continued to discharge longer. The difference in discharge between the two drain sets may have been caused by heterogeneity of hydraulic properties of the subsoil (Stuyt et al., 2000); this would imply that the subsoil of Drain Set 2 is more permeable than that of Drain Set 1. In addition, the approach-flow or entrance resistances of Drain Set 2 may for some reason be smaller than those of Drain Set 1. No data are available to further support these two possible explanations, yet a substantial variability of drain discharges at field scale is not uncommon for Dutch soils that are known to be quite heterogeneous due to irregularly shaped sedimentation patterns (Stuyt, 1992). Figure 4 shows a few soil moisture profiles that represent almost the full range of the moisture profiles of all sampling dates. Figure 4 shows that the average volumetric water content below 10 cm was never smaller than about 0.3 (not even on Day 125 when the ground water level was almost at its deepest level). Such high moisture contents even in summer are typical for clay soils.. The standard deviations based on the 16 sampled profiles generally ranged between 0.02 and 0.06 m3 m–3, which implies that the spatial variability was only moderate.



View larger version (29K):
[in this window]
[in a new window]
  		Fig. 3. Rainfall, ground water level, and drainflow as a function of time at the experimental field. (A) Daily rainfall and daily average ground water level; the shaded area is the range of the depths of the drain pipes. (B) Cumulative rainfall and cumulative drainflow. Time zero is 0000 h on 7 Apr. 1998.

 


View larger version (13K):
[in this window]
[in a new window]
  		Fig. 4. Volume fraction of water as a function of depth as measured at the experimental field at 22, 69, and 125 d after bromide and bentazon application (Day 22 is 29 Apr. 1998, Day 69 is 15 June 1998, and Day 125 is 10 Aug. 1998). The thick line is the average of the 16 sampled soil columns and the thin lines are the average plus or minus one standard deviation of these columns.

 
Close inspection of changes in ground water levels and drainflow rates immediately after heavy rainfall might reveal preferential flow effects. The only drainflow event in spring (0.4 mm in Drain Set 2 between Days 19 and 21) was, therefore, examined in detail. Figure 5 shows that at the start of Day 18 the ground water level was already close to the depth of the drain pipes. In the second half of Day 18, about 12 mm of rain fell, which led to a sharp rise of the ground water level at the end of Day 18. Almost simultaneously the drainflow started. At the end of Day 20, about 8 mm of rain fell. Almost immediately after the first 2 mm of rainfall, the ground water level rose and drainflow started. Total drainflow of the second event in Fig. 5 is only about 5% of the rainfall of Day 20. The very fast response of the ground water level (rise within a few hours after rainfall) is a strong indication of preferential flow. It would be a strong indication of preferential water flow if the drains would flow while the ground water level was deeper than the tile drains. We checked the time series of ground water levels and drainflow events over the whole experimental period but did not find such events. Admittedly, these events are quite unlikely because the drain pipes have a diameter of about 10 cm and are at a 10-m distance from each other: If water would flow through macropores at the drain depth (at about zero suction), only a very small fraction of this macropore water could be collected by the drains.



View larger version (15K):
[in this window]
[in a new window]
  		Fig. 5. (A) Ground water level, (B) rainfall, and (C) drain discharge in Drain Set 2 as a function of time on Days 18 to 21 (i.e., 25 to 28 Apr. 1998) as measured at the experimental field. The dashed lines in (A) indicate the range of the drain depth. Note that the scale of the vertical axes of (B) and (C) differ.

 
Behavior of the Solutes in the Soil Profile
Figures 6 and 7 compare the behavior of the three solutes in soil. Let us first consider the results for bromide, which are somewhat surprising. The amount of bromide in the 120-cm soil profile sampled 1 d after application corresponded reasonably well with the bromide dose of 60 kg ha–1 (Fig. 6A). However, it decreased about 80% in a few weeks time. This decrease could be completely attributed to bromide uptake by the crop; on Day 34 about 50 kg ha–1 was recovered from the crop. Thereafter the amount of bromide in soil increased gradually, reaching a level of about 40 kg ha–1 on Day 167. The amount in the crop decreased strongly from about 50 kg ha–1 on Day 34 to about 10 kg ha–1 on Day 120 (Fig. 6A). As previously described, the crop was cut on 1 September (Day 135) but the chopped crop remnants were not removed from the field. On Day 22 the highest bromide concentrations in soil were found in the top 10 cm (Fig. 7A). On Day 167 (21 September) the peak of the bromide concentration was between depths of 10 and 20 cm. Most of the bromide had leached from the top 120 cm after the following winter (Day 378 is 19 April of next year). Ostensibly, a large fraction of the bromide dose was taken up by the winter wheat in the first weeks after application [similar to results found by Owens et al. (1985) and Kung (1990)]. Thereafter it was gradually released by the crop leading to the increase in bromide amounts in soil between Days 22 and 167. The subsequent decrease between Days 167 and 378 was the result of leaching.



View larger version (15K):
[in this window]
[in a new window]
  		Fig. 6. Amounts of (A) bromide, (B) bentazon, and (C) imidacloprid recovered from soil and the crop as a function of time as measured at the experimental field. Points are the averages and bars are the standard errors. Closed symbols are amounts recovered from the top 120 cm of soil and open symbols are amounts recovered from the crop (only for bromide). Time zero is 0000 h on 7 Apr. 1998. Note that the vertical axis of (B) is logarithmic and that imidacloprid was applied on Day 50. The arrow in (B) indicates that the point is an upper limit.

 


View larger version (29K):
[in this window]
[in a new window]
  		Fig. 7. Concentration profiles of (A) bromide, (B) bentazon, and (C) imidacloprid in soil as measured at the experimental field at different sampling dates. The thick lines are average values and the thin lines are averages minus or plus one standard deviation of the four samples of each layer. Days indicate the period elapsed since the bromide and bentazon application on 7 Apr. 1998. Day 22 is 29 Apr. 1998, Day 69 is 15 June 1998, Day 125 is 10 Aug. 1998, Day 167 is 21 Sept. 1998, Day 239 is 2 Dec. 1998, and Day 378 is 20 Apr. 1999. Note that the horizontal axis of (A) is linear whereas the horizontal axes of (B) and (C) are logarithmic. The shaded area in (A) and the arrows in (B) indicate that the concentration is below the limit of quantification (LOQ).

 
It is known that old plant parts (e.g., leaves, roots, and fruits) tend to release electrolytes because cell membranes become more permeable. Therefore, rain can leach considerable amounts of ions (e.g., potassium) from crops to the soil (Baker and Hall, 1988). This supports our above interpretation. Bromide was intended to be a tracer of water flow. However, the complications in the bromide behavior shown in Fig. 6A seriously restrict the usefulness of the bromide results for this purpose.

The amount of bentazon recovered from the top 120 cm of soil decreased approximately exponentially with time (Fig. 6B). One day after application, 1.4 kg ha–1 of bentazon was recovered, which is about 5% higher than the applied amount of 1.33 kg ha–1. The decline proceeded rapidly; after 22 d only 47% of the applied amount was left; after 69 d, 15%; after 125 d, 5%; and after 239 d, less than 0.4%. Most of the bentazon was found in the top 30 cm of the soil profile after 22 d (Fig. 7B). At the same time, an average concentration in soil of about 2 µg dm–3 was found between a 70- and 90-cm depth (axis was chosen logarithmic to be able to see whether small fractions of the dose penetrated into subsoil). After 125 d bentazon was almost evenly distributed in the soil profile. On Day 239 the average concentration in the 10- to 20-cm layer was as low as 0.8 µg dm–3 and all other layers had concentrations below the LOQ (also 0.8 µg dm–3). This implies that the concentrations in the 30- to 90-cm layer decreased by roughly a factor of 10 between Days 125 and 239. The laboratory half-life of bentazon in the 40- to 70- and 80- to 120-cm layers ranged from about 100 to 600 d (Table 2). Thus, the decline by a factor of 10 is unlikely to be the result of degradation. The crop was harvested on Day 135 so plant uptake cannot have contributed significantly to the decrease either. So the observed decrease in the 30- to 90-cm layer is likely to be the result of leaching.

Bentazon and bromide were applied on the same day and they are both anions with no measurable sorption in this soil. Nevertheless, they behaved differently in the soil profile in the first 22 d: The amount of bromide in soil declined to 20% of the dose whereas the amount of bentazon declined to about 50% of the dose in that period. The bromide amount in soil decreased faster than the bentazon amount although bromide is a conservative substance, which can only be taken up by the crop (bentazon can also be degraded in soil). The shape of the concentration profiles of bromide and bentazon on Day 22 also differed: bentazon concentrations in the 0- to 10- and 10- to 20-cm layers were approximately equal whereas the bromide concentration in the 0- to 10-cm layer was an order of magnitude higher than in the 10- to 20-cm layer. We attribute this difference in behavior to the high plant uptake of bromide as described before.

Figures 6C and 7C show the behavior of the moderately sorbing imidacloprid in the soil profile. The average recovered amount of imidacloprid in the soil (0–120 cm depth) was 0.55 ± 0.09 kg ha–1 two days after application (i.e., on 29 May 1998, Day 52 in Fig. 6C), which is about 79% of the applied amount of 0.7 kg ha–1. Between the end of May and the second measurement in September a decline of more than 50% was observed. The remaining three measurements in Fig. 6C show almost equal values of about 0.25 kg ha–1. So in the subsequent autumn–winter half-year no further decline was measured, probably because it was too cold.

On Day 167 (117 d after application) the highest imidacloprid concentration was found in the top 10 cm (Fig. 7C). The concentration profile on Day 167 is very steep; the average concentration in the 10- to 20-cm layer is about five times lower than that in the 0- to 10-cm layer and the average concentration in the 30- to 120-cm layer is two orders of magnitude lower than that in the 0- to 10-cm layer. On Day 378 imidacloprid had moved some 10- to 20-cm downward; the concentration peak is in the 10- to 20-cm layer. Probably the greatest part of this movement may be attributed to plowing (25–30 cm deep on Day 245). So the movement of the bulk of the imidacloprid was quite slow during the winter period. The imidacloprid concentration profile measured on Day 239 showed average concentrations that were intermediate between those measured after Days 167 and 378 for almost all layers.

All concentration profiles in Fig. 7 were measured after mixing of corresponding layers of four plots (see Materials and Methods), thus reducing the 16 individual samples of each layer to four samples before analysis. Admittedly, analysis of all individual columns might have revealed important characteristics of the leaching process. Thus, the mixing resulted in loss of information at the scale of individual soil columns. According to statistical theory, the standard deviation of the sample mean, {sigma}M, based on a random sample of size n is the following function of the standard deviation of the population, {sigma} (Bhatthacharyya and Johnson, 1977):

[1]

Assuming that the four mixed samples were randomly taken, the mixing of four layers is expected to have reduced the standard deviation of the concentrations in the individual soil columns by a factor of two.

Figures 7A and 7B show that the spatial variability of the concentration profiles of the two mobile substances bromide and bentazon was considerable. The standard deviations based on the four mixed samples were usually in the order of 50% of the mean values. Figure 7C shows that the standard deviations of the imidacloprid concentrations in the 0- to 10- and 10- to 20-cm layers at Day 167 were comparatively small. However, for all other layers the standard deviations were usually again in the order of 50% of the mean values.

The concentration profiles of the three substances in Fig. 7 do not show evidence of preferential transport; no bimodal concentration profiles were found and the concentrations of bentazon and imidacloprid below a 50-cm depth were so low that they could only be visualized via using logarithmic concentration axes. So these measurements of resident concentrations were not suitable for demonstrating preferential transport.

Drain Water Concentrations and Their Consistency with Soil Profile Concentrations
Figure 8 compares the drain water concentrations of all three substances. Bromide concentrations of Drain Set 1 and 2 were remarkably close to each other during the whole experimental period. The highest bromide concentrations were found for the only drainflow event in spring (Day 21) and for the first drainflow event in summer (Day 149). In general the variation in time of the bromide concentrations was small; all concentrations ranged between 2000 and 6000 µg dm–3 (except one outlier around Day 200). So the highest bromide concentrations were found for the drainflow event shown in Fig. 5. Then, Drain Set 2 produced much more discharge than Drain Set 1 (0.4 mm vs. less than 0.005 mm). However, the concentrations for the two sets were remarkably close on Day 21 (Fig. 8A). The high bromide concentrations on Day 21 are clear evidence of preferential transport; the amount of water in the soil profile above the drain depth is of the order of 300 mm (Fig. 4) so matrix fluxes cannot explain the high concentrations after only 56 mm of rainfall. The total amount of bromide that leached via drainflow on Day 21 was 0.035% of the dose for Drain Set 2 and about 1000 times smaller for Drain Set 1 because of the very low drain discharge.



View larger version (16K):
[in this window]
[in a new window]
  		Fig. 8. Concentrations of (A) bromide, (B) bentazon, and (C) imidacloprid in drain water for both drain sets (closed symbols for Set 1 and open symbols for Set 2) as a function of time as measured at the experimental field. Time zero is 0000 h on 7 Apr. 1998 (i.e.. the application date of bromide and bentazon; imidacloprid was applied on Day 50). Note that vertical axes have different ranges and that in (B) the vertical axis has a break.

 
On Day 22 all bromide concentrations in soil below a depth of 20 cm were lower than the detection limit except an average concentration of about 800 µg dm–3 in the 70- to 90-cm layer (Fig. 7A). This corresponds with about 2000 µg dm–3 in the liquid phase. So the bromide concentration of about 5000 µg dm–3 found in drain water on Day 21 (Fig. 8A) was 2.5 times higher than could be found in the liquid phase at the drain depth (i.e., 80–90 cm). Concentrations in the soil system corresponding with 5000 µg dm–3 in the liquid phase on Day 22 were found only in the top 10 cm. This is strong evidence for preferential flow of bromide through the soil profile in the first three weeks after application. Both on Days 167 and 378 the bromide concentration in the 70- to 90-cm layer was about 1500 µg dm–3 (Fig. 7A). This corresponds with about 4000 µg dm–3 in the liquid phase, which is consistent with the drain water concentrations found during the autumn and winter period (Fig. 8A). Hence, only the bromide concentration measured in drain water on Day 21 provided evidence for preferential transport.

The bentazon concentrations of both drain sets were remarkably close to each other at all sampling dates (Fig. 8B). By far the highest bentazon concentrations (about 90 µg dm–3) occurred already on Day 21 (note the break in the vertical axis of Fig. 8B). The bentazon concentrations of the next drainflow event (on Day 149) were about 10 times lower. The high bentazon concentrations on Day 21 are clear evidence for preferential transport (based on the same arguments as described above for bromide). The total amount of bentazon that leached via drainflow on Day 21 was 0.027% of the dose for Drain Set 2 and about 1000 times smaller for Drain Set 1 because of the very low drain discharge. This 0.027% is very close to that found for bromide for Drain Set 2 (0.035%).

The average bentazon concentration of 2 µg dm–3 found between a 70- and 90-cm depth in soil on Day 22 (Fig. 7B) corresponds with 5 µg dm–3 in the liquid phase (the volume fraction of water in this layer was about 0.4). So the bentazon concentration of 90 µg dm–3 found in drain water on Day 21 was about 18 times higher than the average resident concentration in the liquid phase at the drain depth. Concentrations in the soil system exceeding 90 µg dm–3 in the liquid phase were only found in the top 50 cm on Day 22. This is evidence for preferential transport of bentazon to the drains in the first three weeks after application.

Imidacloprid concentrations in both drain sets were mostly close to each other but not around Days 150 and 260 (Fig. 8C). The highest imidacloprid concentrations in drain water occurred in the first drainflow events in September (between Days 149 and 161). Thereafter the concentrations decreased sharply. Note that imidacloprid was applied on Day 50 so it was not yet introduced into the soil system on Day 21 when the bromide and bentazon peaks were measured in the drain water.

The drain water concentration of imidacloprid was on average about 5 µg dm–3 between Days 149 and 161 (Fig. 8C). Assuming sorption equilibrium using a Freundlich coefficient of 0.75 dm3 kg–1 (based on the measured KF of 1.5 dm3 kg–1 at 15°C for the top layer and divided by two because the subsoil contains about two times less organic matter than the topsoil as shown in Table 1), a Freundlich exponent of 0.81, a dry bulk density of 1.5 kg L–1, and a volume fraction of water of 0.35, this corresponds to a concentration of about 20 µg dm–3 in the soil system. Such concentrations were found only in the top 30 cm on Day 167 (Fig. 7C), which indicates that the drain water concentrations of imidacloprid between Days 149 and 161 resulted from preferential transport.

The average imidacloprid concentrations in soil between a 50- to 90-cm depth were below 2 µg dm–3 on Days 167 and 239 (Fig. 7C). Assuming sorption equilibrium and the same parameters as in the previous paragraph, such a concentration in soil corresponds with a concentration in the soil pore water of 0.4 µg dm–3. Most drain water concentrations of imidacloprid were higher than 0.4 µg dm–3 between Days 167 and 239 (Fig. 8C). This is evidence for preferential flow of imidacloprid from the topsoil to drains in this period of about 70 d (mainly October and November). Drain water concentrations of imidacloprid continued to decrease between Days 167 and 239 especially for Drain Set 2 (Fig. 8C). This is at first glance not consistent with the downward movement in the resident concentration profiles in that period (Fig. 7C). A possible explanation is that the 160 mm of rain that fell between Days 149 and 161 effectively leached the readily available part of the imidacloprid in the macropores from the top 10 cm. The remaining imidacloprid in this layer may have contributed less to the macropore flow because it was located deeper in the soil aggregates, thus leading to the sharp decrease in drainflow concentrations.

Thus, evidence was found for preferential flow to drains of bromide and bentazon between Days 1 and 22 and for imidacloprid between Days 149 and 161 and Days 167 and 239. Between Days 1 and 22 and Days 167 and 239, the soil was very wet; the moisture profiles on Days 22, 167, and 239 were the wettest ones of the experimental period (moisture profiles of Days 167 and 239 were very close to the profile of Day 22 shown in Fig. 4). No shrinkage cracks were observed at the soil surface at these sampling dates. This supports the hypothesis that the preferential flow was caused by the permanent macropores observed in the 30- to 100-cm layer. However, between Days 149 and 161 the ground water table rose from a 1.9-m to a 0.8-m depth due to the heavy rain in this period (Fig. 3). It is very likely that shrinkage cracks were present in soil when this rise started. Thus, shrinkage cracks may have been partly responsible for the preferential flow of imidacloprid observed between Days 149 and 161 (note that permanent macropores then also may have contributed to the preferential flow). So there is convincing evidence that permanent macropores caused preferential flow whereas this evidence for the shrinkage cracks is much weaker. Hendriks et al. (1999) studied bromide transport in a clay soil at an experimental field at about a 10-km distance from our field and also found evidence for preferential flow through permanent macropores.

Early breakthrough of solutes in drain water in macroporous clay soils has also been reported in the literature. Bronswijk et al. (1995) and Larsson and Jarvis (1999) found high bromide concentrations in drain water after 25 to 50 mm of rainfall; this is strikingly similar to our results in Fig. 8A and 8B. Thus, the phenomena observed in our field experiment are probably representative for a broad range of clay soils, which makes the data set attractive to use for testing and calibrating pesticide leaching models.

In Dutch pesticide registration, so far only spray drift is considered as a source of contamination of surface water (College Toelating Bestrijdingsmiddelen, 2002). Spray drift percentages of 1% of the dose are used for arable crops, which results in an initial exposure concentration in surface water of about 2 to 3 µg dm–3 (assuming a dose of 1 kg ha–1). Our drain water measurements in Fig. 8B and 8C show higher concentrations for bentazon and imidacloprid. This indicates that for Dutch clay soils preferential flow via drain pipes to surface water may lead to concentrations in surface water that are comparable with or even higher than those resulting from spray drift. Thus, it is not justified to a priori ignore leaching via drain pipes in the risk assessment for surface water in Dutch pesticide registration.

Ground Water Concentrations and Their Consistency with Drain Water Concentrations
Bromide concentrations in ground water at 1.0- to 1.2- and 1.3- to 1.5-m depths increased until about Day 250 (Fig. 9A). Thereafter, the concentrations remained around 3000 µg dm–3 until the end of the experiment. For the deep layer (1.9–2.8 m), low concentrations of bromide around 150 µg dm–3 were found at two times (Days 52 and 125) followed by an increase to about 1000 µg dm–3 on Day 331. We compare now the bromide concentrations in ground water at a 1.0- to 1.2-m depth with those in drain water (Fig. 8A and 9A). The bromide concentration in the first drainflow on Day 21 was about 5000 µg dm–3, which is five times higher than the concentration of about 1000 µg dm–3 found in ground water between depths of 1.0 and 1.2 m on Day 22. As previously described, the concentrations in drainflow at that time were the result of preferential transport. It is not surprising that the concentrations in ground water are lower: Water coming from the preferential flow paths may be diluted in the saturated zone with resident clean water. The concentrations in drainflow and ground water at a 1.0- to 1.2-m depth from about Day 150 onward are more or less equal to each other. As indicated by Fig. 7A, the bulk of the bromide leaches from the top 1 m between Days 167 and 378 while exhibiting only small concentration gradients below a 70-cm depth. Bromide concentrations in the soil profile, drainflow, and ground water are consistent in this period.



View larger version (16K):
[in this window]
[in a new window]
  		Fig. 9. Concentrations of (A) bromide, (B) bentazon, and (C) imidacloprid in ground water collected from different filter depths as a function of time as measured at the experimental field. Time zero is 0000 h on 7 Apr. 1998 (i.e., the application date of bromide and bentazon; imidacloprid was applied on Day 50). Points are the averages and bars are the standard errors in (A) and (B). In (C) the points are individual values and the dashed line indicates the detection limit (points on this line indicate values below the detection limit).

 
By far the highest bentazon concentration in ground water (about 16 µg dm–3) was found at a 1.0- to 1.2-m depth at the first sampling on Day 22 (Fig. 9B). Thereafter the bentazon concentrations in ground water were almost constant with depth and time with the exception of a peak for the shallow ground water around Day 150. The bentazon concentration in ground water of 16 µg dm–3 at a 1.0- to 1.2-m depth on Day 22 is about six times lower than found in the drain water on Day 21. This is consistent with the factor five difference found for the bromide concentrations in the same samples. The lower bentazon concentration in ground water has probably the same cause as the one suggested for bromide: Water leaching downward via preferential flow paths may be diluted with resident clean water after reaching the ground water table.

The imidacloprid concentration in ground water fluctuated usually between 0.1 and 1 µg dm–3 and its spatial variability was considerable as shown in Fig. 9C (note that the vertical axis is chosen logarithmic for a better display of the variability). Imidacloprid was detected in ground water from the 1.3- to 1.5-m layer at a concentration of about 0.09 µg dm–3 on Day 63, which is 13 d after its application on Day 50. In these 13 d about 70 mm of rain fell. A concentration of 0.09 µg dm–3 in a 0.2-m-thick saturated layer corresponds with about 0.01% of the applied amount. Probably this rapid breakthrough is the result of preferential flow (although this is difficult to demonstrate without performing solute transport calculations with the convection–dispersion equation). Comparison of the ground water and drain water concentrations of imidacloprid (Fig. 8C and 9C) shows that around Day 150 the drain water concentrations were about an order of magnitude higher than the concentrations in ground water between depths of 1.0 and 1.2 m. However, at the end of the experimental period the difference between these concentrations had become much smaller.

The ground water concentrations of the three substances in Fig. 9 show strongly different behavior. We consider first bromide and bentazon. At the first ground water sampling (Day 22) their concentrations in the 1.0- to 1.2-m filter were almost proportional to the ratio between their dosages: the bromide concentration was 60 times higher than the bentazon concentration and the bromide dosage was 45 times higher than the bentazon dosage. However, at later times the behavior of both substances in ground water diverged strongly; the bromide concentrations increased to about 3000 µg dm–3 and remained stable at that level whereas the bentazon concentrations decreased strongly. The main cause of this difference is the difference in persistence in the soil–plant system. On Day 167 about 65% of the bromide dose was recovered from the top 1.2 m of soil whereas this percentage was only 5% for bentazon on Day 125. The course of time of the ground water concentrations of imidacloprid is more or less similar to that of bromide; after an initial increase, the concentrations remain more or less constant during the remainder of the experimental period. This similarity is partly the result of the similarity in persistence between the two substances. The concentrations of bromide do not decrease because the bulk of the bromide has not yet leached out of the soil profile. The concentrations of imidacloprid remain constant because there remains an almost constant source of imidacloprid available in the top 30 cm of soil.


   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Drain discharge never occurred while the ground water level was deeper than the tile drains. In the only drainflow event in spring, the ground water level rose within a few hours after rainfall, which indicates preferential flow. For each substance, the highest concentration in drain water was found in the first drainage event after its application, which indicates preferential flow as well. Evidence for preferential flow was found mainly for wet periods in which no shrinkage cracks were observed at the soil surface. This preferential flow is probably caused by permanent macropores observed in the 30- to 100-cm layer. At the time of the first drainage event, the drain water concentration of each substance was about an order of magnitude higher than its ground water concentration at a 100- to 120-cm depth. The cause of this difference is probably that water flowing to the tile drains is less diluted by clean water in the saturated zone than water in the saturated zone below the tile drains.


   ACKNOWLEDGMENTS
 
We would like to thank the Fundação Coordenação de Aperfeiçoamento de Pessoal de Nível Superior-CAPES (Brazil) and the Dutch Ministry for Agriculture, Nature and Food Safety for funding.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES