| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Laboratory of Soil Science, INRA, 2 place Viala, 34060 Montpellier Cedex 1, France
Corresponding author (louchart{at}ensam.inra.fr)
Received for publication February 15, 2000.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
The main objective of this study was therefore to assess and to link pesticide runoff at the field and watershed scales under Mediterranean climate conditions in a sedimentary agricultural watershed, typical of the vineyard area in southern France. In a previous work (Lennartz et al., 1997), we analyzed the main processes governing herbicide dissipation and runoff at the field scale during the years 1994 and 1995. In particular, we showed a high removal potential of chemicals and herbicides by overland flow due to (i) the intensity of Mediterranean storm events, which induce erosion and intense runoff and (ii) the dry and hot summer conditions, which slow down the pesticide degradation in the topsoil. In this paper, the specific objectives were to: (i) determine which processes are involved in transport of herbicides into surface waters from the field to the watershed and (ii) quantify the influence of these processes and of the different tillage practices on total loads of herbicide at the watershed scale. The herbicides diuron and simazine were chosen for this study, because they are the most widely used herbicides controlling weeds in vineyards in the studied watershed.
|
MATERIAL AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
|
|
During the study period, vineyards covered 65% of the area of the watershed, whereas cereals, alfalfa, and market gardening occupied 6%, and fallow and scrubland represented about 25%. The fields were not irrigated.
A survey identified two main soil treatments currently in use in vineyards for controlling weeds. In one, herbicides are only applied along the rows and the soil is tilled with a rotovator between the vine rows one to three times during the growing period (row weeding). In the other, herbicides are applied over the whole vineyard without any tillage (complete weeding). In both treatments, herbicide application is performed between March and May for most fields. In 1997, in the Roujan watershed, the first treatment was applied on 27% of the area cropped with vines, whereas the second was applied on the remaining 73%.
In order to study the transport processes of herbicides as influenced by the cultivation techniques, we chose two field sites in the watershed, representing both treatments. The study sites will be identified hereafter as the no-till and the tilled site. Both fields were located on the terraces. The main characteristics of each field plot are given in Table 1. Detailed characteristics about soil and planting of vine stocks of both fields can be found elsewhere (Lennartz et al., 1997). In 1997, the tilled field was tilled twice on 16 May and 1 July.
Hydrological Monitoring
As shown in Fig. 1, rainfall was measured with tipping bucket raingauges at three locations in the watershed: on both study fields and on the plateau. The rainfall data presented hereafter at the watershed scale is the arithmetic mean of the data measured by the three raingauges.
Runoff discharge at the outlet of both fields and of the watershed was continuously recorded by means of Venturi flumes (Techniflow, Echirolles, France). At the two field outlets, maximum observed runoff discharge within the flume was 50 L s-1 and at the outlet of the watershed the maximum value was 1500 L s-1.
The piezometric head at ca. 2 m depth in the depression of the watershed was monitored with a piezometer located near the outlet of the watershed at a distance of 5 m from the main ditch (Fig. 1). The piezometer was 235 cm deep and consisted of a 5-cm-diam. rigid plastic tube that was perforated over 0.30 cm at its base. It was installed in the middle of a 6.5-cm auger hole. The space between the base of the tube and the soil was filled with gravel, and the equivalent space at the top was filled with bentonite to prevent water from infiltrating down the side of the piezometer.
Application Rate and Soil Sampling
In 1997, the herbicides were applied as suspensions by using the commercial product Trevi10 (Calliope S.A., Noguères, France) on both tilled and no-till fields at a 12 L ha-1 recommended rate, which corresponds to rates of 2 and 1 kg a.i. ha-1 of diuron and simazine, respectively. The application dates were 2 April for the tilled field and 7 April for the no-till field.
To determine diuron and simazine persistence on both fields, sampling of the soil surface (0–2 cm) was performed once prior to chemical application, at the date of application, and then every 2 wk during the first month after application and every month for the following 3 mo. From the second month after application onward, sampling was extended also to two sublayers, 2 to 5 and 5 to 15 cm. For estimating unbiased field average values of herbicide concentrations in soil, spatial sampling was performed at each sampling date following a stratified random scheme that took into account the row-cropped structure of the vineyards. The full details of the sampling design are described by Lennartz et al. (1997).
Water Sampling
At the field and watershed outlets, water sampling was performed by two automated pumping samplers (Sigma Model 800SL; American Sigma, Medina, NY) with a capacity of 24 glass bottles: one was driven by the water table height within the weir and the other sampled at regular time intervals when field runoff started. With this protocol it was possible to ensure a satisfactory sampling both of the whole runoff event and of its rising limb. Because the duration of runoff events was different between the field and watershed scales, the sampling time steps of the second sampler were set to 20 min at the field outlets and to 40 min at the watershed outlet. In some events a malfunction only allowed the determination of one average concentration. Water samples from baseflow were taken manually at the outlet of the watershed between storm events.
Sampling of the ground water in the depression of the watershed was performed monthly or when cumulative rainfall exceeded 50 mm since last sampling. It was done in the piezometer described previously. The ground water samples were taken with a manual pump and stored in a glass bottle; their volume was between 0.5 to 1 L.
Analytical Methods
All water and soil samples were stored frozen at -20°C until analysis. Water samples (200 mL to 1 L) were filtered at 0.45 µm (MAGNA Nylon; Micron Separations, Westboro, MA). Separation was followed by a liquid–liquid extraction of the aqueous phase with 30 mL dichloromethane during 30 min shaking. Solvent extract was divided into two aliquots that were placed in round-bottom roto-evaporation flasks. Samples were evaporated to dryness. Soil samples (50 g) were sieved to 5 mm prior to chemical extraction by shaking with 100 mL methanol for 12 h. After decanting and filtration, two aliquots of 20 mL each were evaporated to dryness at 40°C. Subsequent analyses were the same for both soil and water samples. Simazine was analyzed by capillary gas chromatograph (GC) using a Varian (Walnut Creek, CA) Model 3400 CX equipped with a N-P detector. The applied internal standard, terbutryn, revealed a recovery rate of the method of 88%. A Varian high performance liquid chromatography (HPLC) system was used for the determination of the diuron concentration. Recovery rate of the method was 94%. More details about extraction, GC, and HPLC conditions can be found elsewhere (Lennartz et al., 1997). Detection limits of the procedure for both diuron and simazine were 0.05 µg L-1 and 1 µg kg-1 for water samples and soil samples, respectively.
All results presented hereafter are not corrected for the recovery rate. Separated sediment was not analyzed for herbicide content because the time period between automatic sampling and collection of samples (average of 16 h) allowed equilibration between concentration in the liquid and solid phases (sediment). Furthermore, measured sediment mass was <2% of total sample weight in >96% of the analyses. Our underestimation of herbicide load because the solid phase was not analyzed was calculated from sediment load and the measured value of linear adsorption coefficient, and found to be <6%. Moreover, many previous works have shown that most herbicides are essentially transported in the liquid phase (Baker et al., 1978, 1982; Baker and Johnson, 1979; Wu et al., 1983; Klaine et al., 1988; Gouy, 1993). Thus, runoff pesticide loads presented hereafter are based on results from analyses of water samples only.
Estimating the Amount of Herbicides Applied to the Watershed
A survey carried out with the vine growers of the watershed allowed us to estimate the application rate and the date of application of diuron and simazine on 127 of 131 vineyard fields of the watershed. For the four remaining fields for which the farmer could not be surveyed, it was assumed that diuron and simazine were applied according to the application rates recommended by the manufacturer, namely 2 and 1 kg a.i. ha-1, respectively. To estimate the proportion of the watershed area that was treated, we assumed that the fields on which chemical weeding was restricted to the vine rows were sprayed over only one-third of their area.
In accordance with the French regulations controlling the use of herbicides in fields, simazine was always applied in combination with diuron, whereas on some fields diuron was applied alone. This implies that the spatial and temporal patterns of application were different for diuron and simazine. The survey indicated that at the watershed scale 87 and 57% of the area cropped with vines was treated with diuron and simazine, respectively (Table 1).
In 1997, the period of application of herbicides lasted from 1 March to 15 May (Table 1). Almost 80% of the annual amounts of applied diuron and simazine were sprayed on the fields in March, whereas the remaining 20% were sprayed in April and May.
Estimating Herbicide Losses at the Field and Watershed Outlets
Herbicide loads were determined by integrating concentrations with water flux over the sample-time intervals. During periods of storm flow, the automatic sampling devices provided a satisfactory temporal sampling resolution. But, during periods of baseflow, only a few samples were taken. Consequently, to improve the temporal resolution of the sampling we assumed in the calculations that the herbicide concentrations measured in the ground water at the vicinity of the watershed outlet (2 June and 27 Nov. 1997; 22 Jan. 1998) were also representative of those in baseflow. As will be shown hereafter, this assumption is justified by the fact that baseflow arises from the drainage of the ground water. Finally, it is important to notice that for the purpose of the calculations, herbicide concentrations below the detection limit of 0.05 µg L-1 were set equal to zero. This method can induce a small bias if most of the samples exhibit concentrations below the detection limit, but this was not the case.
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
|
The comparison of the average herbicide concentrations in the surface layers of the no-till and tilled fields shows that the loads in the former are larger than in the latter, which was expected because the tilled field was only treated along the rows. But, in contrast, no significant difference appeared between the two fields in terms of decay pattern.
Patterns of Herbicide Transport by Overland Flow at the Field Scale
Characteristics of Runoff Events
The hydrological data of all sampled runoff events are plotted in Fig. 3. For each event a runoff coefficient (RC) was computed as the ratio between the runoff amount and the rainfall depth, both expressed in millimeters. The runoff coefficients allow us to compare runoff intensity between different fields that did not receive the same amount of rainfall.
|
|
|
Herbicide concentrations in runoff water and herbicide loads from the tilled field were smaller on average than those from the no-till field. This difference arises from the smaller herbicide concentration of the topsoil and was accentuated by the smaller runoff volumes at the tilled field in comparison with those at the no-till field.
Lastly, in most runoff events, diuron concentrations were larger than those of simazine. This is possible because either simazine concentrations are smaller in the soil (Fig. 2) or simazine has a larger adsorption capacity on soil than diuron (Baer, 1996).
Herbicide Transport from Field to Watershed Scale
Watershed Hydrology
Figure 5a–c shows that the watershed hydrology is clearly event-dominated. Similar to observations in previous years (Voltz et al., 1997, 1998), there were four periods with different hydrological characteristics in 1997.
|
The second period corresponds to the abrupt and short climatic transition that often happens in the Mediterranean area at the end of the dry summer conditions when the first heavy rainfalls occur. In 1997 it happened in early November. The large volume and intensity of the rainfall cause large overland flow on the fields, and thereby allow both intense discharge at the watershed outlet and complete recharge of the ground water by seepage from the ditches. Hydrochemical tracing performed by Ribolzi et al. (2000) during the first rainfall events after the summer of 1994 confirmed that up to 80% of the discharge at the watershed outlet during the intense runoff events was due to overland flow generated at the field scale.
The third period includes autumn and part of the winter, during which rainfall is usually maximum and the ground waters of the watershed are high. In this situation, the network of ditches drains the ground water, which produces a permanent baseflow. But during the rainfall events, the watershed discharge is still dominated by direct overland flow. In effect, although the intensity of rainfall is, in general, less than during the second period, intense overland flow still occurs on the fields, because the infiltration capacity of the soils of the watershed is smallest during this third period. This is because the soils remain mostly wet and even the tilled fields have formed crusts with a small infiltration capacity.
The fourth period corresponds to the end of winter and to the beginning of spring. During this period rainfall events are not, in general, as frequent as in the third period. Consequently the discharge at the outlet of the watershed is mostly due to baseflow and decreases progressively from its maximal value (2 mm h-1) to nothing.
During 1997 the total watershed discharge amounted to 205 mm, as can be seen in Table 2. The amounts of discharge that occurred by storm flow and baseflow were estimated as follows. Storm flow was assumed to start when the watershed discharge increased substantially in relation to a rainfall event, and was assumed to end either at the time of the point of inflection on the recession limb of the hydrograph or, if it occurred earlier, at the time at which discharge returned to its initial value. The amount of baseflow was estimated by the difference between annual discharge and storm flow. Table 2 shows the estimated contributions of storm flow and baseflow. Because storm flow is mainly generated by overland flow exiting the fields, it follows that less than 81 mm of field overland flow reached directly the watershed outlet by the network of ditches. Given the amounts of observed overland flow at the field scale, this means that a large part of the field overland flow reinfiltrated during the storm events in the ground water of the watershed through the bottom of the ditches and exited the watershed as baseflow between the storm events.
Herbicide Concentrations
The evolution of the measured concentrations of diuron and simazine in runoff water at the outlet of the watershed and in the ground water is plotted in Fig. 5 for the whole monitoring season. The concentrations in diuron and simazine in storm flow water at the watershed outlet decreased regularly during 1997, similarly to what was observed at the field scale. This suggests that the herbicide concentrations at the watershed outlet during the runoff events are closely linked to those at the field outlets. However, we must distinguish the behavior of diuron and simazine because they are different.
The concentrations of diuron in storm flow at the watershed outlet were large, up to 532 µg L-1 during the first runoff event after application. They were closer to herbicide concentrations observed at the outlet of small research plots (e.g., Hall et al., 1991; Lennartz et al., 1997; Troiano and Garretson, 1998) than to those observed in runoff water at the outlet of small or medium-size watersheds (e.g., Matthiessen et al., 1992; Williams et al., 1991, 1995; Richards and Baker, 1993; Garmouna et al., 1997, 1998). In effect, for all runoff events, maximum and mean diuron concentrations measured at the outlet of the watershed were always close to those measured at the no-till field and clearly larger than those at the tilled field (Fig. 6). This means that no significant dilution of the diuron concentrations occurred between the field and watershed scale. There are two major implications. First, it confirms that overland flow is the main contributor to watershed discharge during storms, because concentrations of diuron in baseflow are much smaller than those in overland flow, as will be shown below. Second, it suggests that the contribution to watershed discharge by overland flow generated by the fields that were not sprayed with diuron was minor, although they represented 43% of the watershed area. In effect, half of the nontreated fields were fallow or scrubland with large infiltration capacity and little overland flow.
|
A close examination of diuron and simazine concentrations in baseflow and in the ground water of the depression (Fig. 5d,e) shows two points. First, diuron and simazine concentrations in baseflow samples were mostly similar to those in the ground water samples. This is consistent with the fact that baseflow arises from the drainage of the ground water. Other studies already showed a similar discharge of herbicides from the alluvial aquifer to the river (Frank and Sirons, 1979; Wang and Squillace, 1994; Fisher et al., 1995; Garmouna et al., 1997). Second, herbicide concentrations in the ground water were one or more orders of magnitude less than those in storm flow up to November 1997. Moreover, their fluctuations can be considered as small with regard to the concentration fluctuations of storm flow water. In addition, no significant changes in the herbicide concentrations of the ground water happened when the ground water was replenished by the storm events of 1 and 5 June and 3 and 4 Nov. 1997. To explain this relative stability of the concentrations in the ground water there are two possible hypotheses. One is that the recharge of the ground water is not only due to highly contaminated overland flow water that reinfiltrated by seepage through the ditches, but also to soil drainage water, which is less contaminated by herbicides. Another is that adsorption and degradation occur, which limit the variation of the herbicide concentrations in the aqueous phase of the ground water. At this stage it is not possible to speculate on the relative importance of these two processes. However, the estimation of the herbicide losses by baseflow, presented below, will give us some insight into this problem.
Herbicide Losses
Losses of diuron and simazine per storm event, as well as the cumulative losses during the storm events, are plotted in Fig. 4. The losses at the watershed scale did not decrease from one event to the other as was observed at the field scale. The largest losses were recorded during the first three storm events due to the large herbicide concentrations in storm flow: in particular, the losses during the second runoff event amounted to 28 and 45% of the total seasonal losses of diuron and simazine, respectively, that occurred during storm flows. During the storm event of 3 and 4 November, 6 mo after the application period, significant losses occurred again, although herbicide concentrations were not larger than 1 µg L-1. They were caused by the large runoff volume, which represented 37% of the seasonal storm flow. Thus, in the conditions of this study site, important herbicide losses in runoff water can be observed throughout the year due to either large herbicide concentrations or large runoff volumes. This is in contrast with observations in temperate climate conditions: Matthiessen et al. (1992) reported that simazine losses from a 180-ha watershed decreased from the first runoff event after application onward, and Jaynes et al. (1999) measured in embedded watersheds of 9 to 5130 ha that atrazine was carried out only during 4 mo after application.
The estimated total seasonal losses, including losses by storm flow and baseflow, amounted to 0.52% for diuron and 0.24% for simazine of the applied quantity (Table 2). These losses were similar to those reported in previous studies for simazine (Gomme et al., 1991; Williams et al., 1995; Garmouna et al., 1997) or for other herbicides, mainly atrazine (Fisher et al., 1995; Ng et al., 1995). The estimated values of the specific losses due to storm flow and to baseflow indicate that baseflow only contributed to 4% of the total losses for diuron, and to 17% for simazine. This shows two points. First, herbicide losses by baseflow appear to be of minor importance in comparison with losses by storm flow, despite the large contribution of baseflow to the annual runoff volume. Of course, this comes from the large difference in herbicide concentrations of storm flow and baseflow waters. Second, the small losses by baseflow suggest that adsorption and/or degradation processes occur in the aquifer. In effect, we showed previously that a large part of the field overland flow reinfiltrated in the ground water and the remaining contributed to storm flow at the watershed outlet. Consequently, because overland flow is the main contributor to water and herbicide discharge at the watershed outlet during storm periods, we can assume that at least the same amount of herbicide was transported to the ground water than to the watershed outlet during the storm periods. Thus, if there were no adsorption and degradation processes in the aquifer, herbicide losses by baseflow, which drains the ground water, should have been of the same order of magnitude as those by storm flow on the annual scale.
The discrepancy between diuron and simazine results may be explained by the fact that simazine concentrations in storm flow were one order of magnitude less than those of diuron during a large part of the monitoring season, whereas the concentrations of both compounds in baseflow were always similar.
Finally, the comparison of seasonal losses between the field and watershed scales shows that the losses, as expressed in percentages of applied amounts, are significantly smaller than those at the field scale (Table 2). This was expected because a part of the herbicides extracted by overland flow at the field scale reinfiltrated to the ground water and were not transported again by baseflow.
|
SUMMARY AND CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
(i) A prolonged persistence of a few weeks was observed for diuron and simazine after application in the soil surface layers as compared with the data reported in literature and obtained under temperate climate. This may be related to the inhibiting effect of the very dry summer conditions in the Mediterranean zone on pesticide degradation. Thus, a potential source of herbicides for runoff water remained throughout the growing season.
(ii) At the field scale, losses of herbicides occurred during storm events and were due to intense overland flow caused by the large rainfall intensities of the Mediterranean climate; more than 84% of the annual loads of diuron and 94% of those of simazine were removed from the fields by only four storm events, which corresponded to less than 10% of the annual runoff volume. Maximum concentrations, amounting to 800 µg L-1 for diuron and 580 µg L-1 for simazine, were observed during the first storm event after application. Thereafter, the concentrations decreased from one event to the other, but remained for several months above 1 µg L-1. Tillage substantially reduced herbicide losses in comparison with no-tillage because it restricted runoff by increasing the infiltration capacity of the soil.
(iii) At the watershed scale, most herbicide losses also occurred during the storm events and were caused by the fast transmission through the network of ditches of a minor part of the overland flow exiting the fields. Consequently, for diuron, which was sprayed over most vineyards, concentrations in runoff water at the watershed scale were close to those at the field scale and were therefore greater than herbicide concentrations observed at the outlet of other watersheds (Garmouna et al., 1997). For simazine, which was only sprayed over slightly more than half of the vineyards, dilution by water coming from nontreated fields decreased its concentrations at the watershed outlet.
(iv) A large decrease in herbicide losses by runoff water was observed between the field and watershed scale. It can be explained by the fact that a major part of the herbicide losses at the field scale reinfiltrated to the ground water of the watershed by seepage through the ditches. But, surprisingly, the herbicide concentrations of the ground water, and in baseflow as well, remained small and rather stable throughout the growing season. This leads us to hypothesize that important degradation and retention processes of herbicides occur in the aquifer of the ground water.
In conclusion, it seems that the climatic characteristics of the Mediterranean environment combined with the agricultural practices used in vineyards lead to an intense, but event-based, contamination of surface waters by herbicides during almost the whole growing season both at the field and watershed scales. The main pathways by which herbicides are transmitted from the soil surface to the watershed outlet are overland flow at the field scale and concentrated flow in the network of ditches. In the conditions of our study area, the results obtained indicate two possible ways for minimizing herbicide losses by runoff water. The most straightforward way is to limit chemical weed control by tilling the soil between the vine rows and limiting herbicide application to the rows. This restricts both the amount of herbicide applied and the amount of overland flow at the field scale, and in turn it will also restrict the herbicide losses at the watershed scale because they are mainly caused by overland flow. A complementary way is to retard field to stream transmission of the contaminated overland flow by encouraging infiltration of both runoff and herbicides in the ditches. Further studies are necessary to better understand how herbicides move with runoff water in the ditches, and which design of the ditches would maximize infiltration and retardation of herbicide fluxes.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIAL AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Larose, G. C. Heathman, L. D. Norton, and B. Engel Hydrologic and Atrazine Simulation of the Cedar Creek Watershed Using the SWAT Model J. Environ. Qual., March 1, 2007; 36(2): 521 - 531. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. E. Rupp, R. E. Peachey, K. L. Warren, and J. S. Selker Diuron in Surface Runoff and Tile Drainage from Two Grass-Seed Fields J. Environ. Qual., January 5, 2006; 35(1): 303 - 311. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Crop Science | |||
| Journal of Natural Resources and Life Sciences Education |
Vadose Zone Journal | ||||
| Soil Science Society of America Journal | Journal of Plant Registrations | The Plant Genome | |||
