| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Agriculture and Agri-Food Canada, Research Centre, Lethbridge, AB, Canada T1J 4B1
b National Water Research Institute, Saskatoon, SK, Canada S7N 3H5
* Corresponding author (allan.cessna{at}ec.gc.ca).
Received for publication December 17, 2002.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Prairie farm dugouts generally range in size from approximately 2 to 5 million L. The surrounding drainage areas from which they receive surface runoff water can vary in size from less than 10 ha to greater than 1000 ha. Dugouts are generally located on or immediately adjacent to tilled farmland. As such, they represent a worst-case scenario with respect to contamination with pesticides applied in their immediate vicinity.
It is well established that pesticides can be transported from treated agricultural land in either snowmelt (Nicholaichuk and Grover, 1983; Waite et al., 1992) or rainfall (Wauchope, 1978) runoff. Since surface runoff is the principal source of water for the majority of prairie dugouts, the presence of pesticides in dugout waters would not be unexpected. However, in spite of their frequency of occurrence across the prairie agricultural landscape and the multifunctional purpose for which these water bodies are used, few studies have reported pesticide levels in dugouts.
The dissipation of some pesticides has been studied in farm ponds and dugouts: atrazine [6-chloro-N-ethyl-N'-(1-methylethyl)-1,3,5-triazine-2,4-diamine] and carbofuran (2,3-dihydro-2,2-dimethyl-7-benzofuranyl methylcarbamate) in farm ponds in Kansas (Klaassen and Kadoum, 1979), simazine (6-chloro-N,N'-diethyl-1,3,5-triazine-2,4-diamine) in Missouri farm ponds (Mauck et al., 1976), 2,4-D [(2,4-dichlorophenoxy)acetic acid] in farm ponds in Florida and Georgia (Schultz and Harman, 1974), trifluralin [2,6-dinitro-N,N-dipropyl-4-(trifluoromethyl)benzenamine] in an Alberta dugout (Fox et al., 1991), and cypermethrin [cyano(3-phenoxyphenyl)methyl-3-(2,2-dichloroethenyl)-2,2-dimethylcyclopropanecarboxylate] in an Alabama pond (Hadfield et al., 1993). Frank et al. (1990) detected pesticides in 63% of water samples from 212 farm ponds in Ontario. Of the 29 pesticides detected, the most frequently detected was atrazine followed by 2,4-D, simazine, PCP (pentachlorophenol), dichlorprop [(±)-2-(2,4-dichlorophenoxy)propanoic acid], and endosulfan (6,7,8,9,10,10-hexachloro-1,5,5a,6,9,9a-hexahydro-6,9-methano-2,4,3-benzodioxathiepine 3-oxide). Recently, 112 farm dugouts were monitored in Alberta for herbicide content as part of the Farmstead Water Quality Survey (Canada-Alberta Environmentally Sustainable Agriculture Agreement, 1996). Based on a single sampling during August 1994, herbicide residues were detected in 47% of on-farm surface (dugouts and other sources) water supplies. Herbicides most commonly detected included 2,4-D, MCPA [(4-chloro-2-methylphenoxy)acetic acid], bromoxynil (3,5-dibromo-4-hydroxybenzonitrile), and dicamba (3,6-dichloro-2-methoxybenzoic acid).
Grover et al. (1997) monitored, over two years (1987 and 1988), seven herbicides commonly used in prairie crop production in 21 Saskatchewan farm dugouts situated within four major soil zones. Water samples were collected in the spring following snowmelt, in mid-July when herbicide application is normally completed, and in the fall before freeze up. Herbicides were detected in all dugouts and, in general, more than one herbicide was detected at each sampling time. The decreasing frequency of detection in these dugouts was 2,4-D > diclofop [(±)-2-[4-(2,4-dichlorophenoxy)phenoxy]propanoic acid] > bromoxynil > MCPA > triallate [S-(2,2,3-trichloro-2-propenyl) bis(1-methylethyl)carbamothioate] > dicamba > trifluralin. Maximum concentrations tended to be seasonal and the frequency of detection generally reflected both herbicide use patterns within the vicinities of the dugouts and the environmental stability (field half-lives) of the herbicides. The authors noted that the very high frequency of detection of 2,4-D, irrespective of site or sampling time, most likely reflected its pervasive use in each of the four soil zones during the previous 40 to 50 yr. The authors also suggested that median herbicide concentrations detected in these 21 dugouts may be an indicator of the general level of contamination of dugout waters by herbicides within the prairie region.
The present study investigates more fully the seasonal variation of herbicide concentrations in prairie dugout waters suggested by Grover et al. (1997). Three prairie dugouts were monitored for herbicide content during the 1995 to 1997 growing seasons. The watersheds surrounding the dugouts were delineated and farmers were surveyed to determine herbicide use within each watershed so that herbicide content in the dugout waters could be related to crop production practices.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
The site was part of a mixed farming operation and the dugout, which was not aerated, was used for livestock watering. Constructed in 1986, the dugout had a capacity of 1.3 million L and a water depth of 2.73 m when full. A conservation tillage management system, which minimized summer fallow as well as cultivation before seeding, was in use at the site. Soil and water conservation practices included snow trapping by stubble and shelterbelts. A 30.3-ha watershed drained into the dugout.
Craik
This site is situated in the Dark Brown soil zone approximately 6 km southeast of Craik, Saskatchewan. The dominant soil association is Weyburn (Saskatchewan Soil Survey, 1992) with the most common profiles being Weyburn Orthic Dark Brown Chernozem and Weyburn Calcareous Dark Brown Chernozem soils. The texture of the soil is clay loam. Slopes at the site vary from 3 to 6%. The 30-yr average annual rainfall for the region is 268 mm (Environment Canada, 1993).
Situated in a grain farming operation, the 5.2-million-L (3.50-m-deep) dugout was used as a source of potable and household water by the farm family. The dugout was constructed in 1983 and was aerated only 2 h per day by a linear diffuser positioned centrally on the bottom of the dugout. The dugout watershed was conventionally tilled and summer fallow was included in the crop rotation. A range of cereal and oilseed crops were grown. The area draining into the dugout is approximately 10 ha.
Tisdale
The third site is located in the Dark Gray soil zone approximately 12 km northeast of Tisdale, Saskatchewan and contains soils of the Tisdale and Eldersley associations (Saskatchewan Soil Survey, 1987). The dominant profiles are Tisdale Orthic Dark Gray Chernozems and Eldersley Dark Grey Luvisols. The soil texture is silty clay to clay. The 30-yr average annual rainfall for the region is 312 mm (Environment Canada, 1993).
The dugout, constructed in 1975, was situated within a grain production operation and was also used as a potable and household water supply. The 3.2-million-L dugout was 3.47 m deep and continuously aerated via a hose equipped with a check valve, which was positioned centrally in the dugout and at a 1-m depth. Pulse, cereal, and oilseed crops were grown using conventional tillage and summer fallow was included in the crop rotation. A 7.3-ha watershed drained into the dugout but, because of the very gentle sloping of the landscape (1–3%), the actual area draining into this dugout was difficult to define for major runoff events such as spring snowmelt.
Water Sampling
1995 Growing Season
Winter Samples. Before snowmelt, baseline water samples were collected in mid-February through a hole drilled in the ice at the center of each dugout. At Craik and Tisdale, samples were collected 1 m below the ice and 1 m from the bottom of the dugout. At Lucky Lake, a single sample was collected at half water depth.
Samples Collected during Snowmelt Runoff.
During snowmelt runoff, no water samples were collected at Lucky Lake and Craik. At Tisdale, because of the indeterminate size of the watershed during major runoff events, this dugout generally overflows during snowmelt with excess water flowing across a 25-m grassed area before passing under a road via a culvert. Using a solar-powered automated water sampler (Streamline 800 SL portable liquid sampler equipped with four 2-L glass sample jars; American Sigma, Loveland, CO) with its intake positioned at the downstream end of the culvert, daily samples of the overflowing water were collected from 21 April until 3 May such that a 40-mL subsample was collected every hour over a 24-h period. The composite samples were then transferred to 1-L amber glass bottles, which were capped with Teflon-lined caps.
Spring to Fall Samples.
Following snowmelt, grab samples were collected at approximately one-month intervals between early May and the third week in October from the center of each dugout at points 1 m above the bottom and 1 m below the surface of the water.
To sample at the two depths, a rope, with two floats attached, was strung across each dugout such that the floats were located centrally in the dugouts. Two Teflon tubes (6.5-mm i.d.) were secured along the length of the rope and then suspended from the floats with weights such that the inlet of one tube was approximately 1 m from the surface (shallow) and the inlet of the other tube was approximately 1 m from the bottom (deep) of the dugout. During sample collection, the outlet of the Teflon tubing was connected to a 1-L suction flask and the system evacuated with a hand-held vacuum pump. After approximately 200 mL of dugout water was collected and discarded (to flush out the tubing), a 1-L sample was collected and immediately transferred to a 1-L amber glass bottle and capped with a Teflon-lined cap.
1996 Growing Season
Samples Collected during Snowmelt Runoff.
A single grab sample from water on top of the ice of the Lucky Lake and Craik dugouts was collected directly into a 1-L amber glass bottle on 15 March during a brief period of snowmelt. Snowmelt runoff at these sites resumed at the end of March and was complete by the end of April. Grab samples from open water near the edge of each dugout were similarly collected at an approximately 0.3-m depth, beginning 5 April and subsequently at 4- to 8-d intervals until the end of April. At Tisdale, samples of overflow water were collected as described previously except that the sampler intake was positioned on the bed of an H-flume that had been installed during October 1995 between the outlet of the dugout and the culvert under the adjacent road. Samples were collected 10 to 14 April but, because of low flows, daily grab samples were taken from the H-flume on 15 to 21 April.
Spring to Fall Samples.
After snowmelt, automated water samplers were installed on all three dugouts with the sampler intakes centrally positioned 1 m below the surface. Since maximum dugout depths varied from 2 to 3 m, the sampler intakes were near half depth. The automated samplers were programmed such that 135-mL subsamples were collected every 2 h over a 5-d period into four 2-L glass sampling bottles. Samples were collected from mid-May to early October.
1997 Growing Season
Samples Collected during Snowmelt Runoff.
At Lucky Lake and Craik, grab samples were collected every 3 to 7 d during runoff (late March to early April) as described previously. At Tisdale, there was not sufficient runoff to cause the dugout to overflow. Consequently, daily grab samples were similarly collected during runoff (17–27 April).
Spring to Fall Samples.
Grab samples were then collected at monthly intervals from late April (Lucky Lake and Craik) or late May (Tisdale) until the end of September.
Sample Handling
Each 1-L grab sample or daily composite sample collected using an automated sampler was stabilized by the addition of 5 mL of concentrated sulfuric acid. In the case of the 5-d composite samples, the four 2-L sampling jars were capped and shaken and a subsample taken by pouring a proportional volume from each jar into a 1-L amber glass bottle for a total volume of 1 L. This sample was then similarly stabilized and the bottle capped with Teflon-lined cap. The 1-L samples were transported to the Agriculture and Agri-Food Canada Research Station, Regina, SK (1995 and 1996) or the National Water Research Institute, Saskatoon, SK (1997) and maintained at 4°C until extraction.
Farmer Survey
The farmers of the Lucky Lake, Craik, and Tisdale study sites were initially surveyed at the end of the growing season and asked to provide the following information: crops grown, herbicides applied, rates (g ha–1) and methods of application, and area (ha) treated with each herbicide. When it became evident, after the 1995 growing season, that herbicides other than those monitored were being applied, the farmers were then surveyed in the spring so that additional herbicides could be included in the multiresidue method.
Sample Analysis
Ten herbicides, commonly used in prairie crop production, were monitored in all water samples from all three sites. Three herbicides were neutral: ethalfluralin [(N-ethyl-N-(2-methyl-2-propenyl)-2,6-dinitro-4-(trifluoromethyl)benzenamine], trifluralin, and triallate. The remaining seven herbicides were acidic: clopyralid (3,6-dichloro-2-pyridinecarboxylic acid), dicamba, mecoprop [(±)-2-(4-chloro-2-methylphenoxy)propanoic acid], MCPA, bromoxynil, 2,4-D, and diclofop (Table 1). As a result of the spring farmer surveys, three additional herbicides were included in the multiresidue method but were monitored only at the sites at which they were applied. Dichlorprop and fenoxaprop {(±)-2-[4-[6-chloro-2-benzoxazolyl)oxy]phenoxy]propanoic acid} were applied in 1995 and 1996 at Tisdale and Craik (Table 2), respectively, and were monitored in the corresponding 1996 and 1997 snowmelt runoff samples and 1996 dugout samples. Metribuzin [4-amino-6-(1,1-dimethylethyl)-3-(methylthio)-1,2,4-triazin-5(4H)-one] was applied at Lucky Lake in 1996 and was monitored in the 1996 dugout water samples and the 1997 snowmelt runoff samples.
|
|
Extraction, Methylation, and Florisil Column Cleanup
The water samples were extracted as described previously (Cessna et al., 1985) with the following modifications: a 500-mL rather than a 1-L sample was extracted and all extraction solvent and reagent volumes were reduced proportionally. For water samples monitored for metribuzin, the neutral herbicides were extracted with methylene chloride rather than hexane and, following elution of trifluralin, ethalfluralin, and triallate from the Florisil cleanup column with 0.5% acetone in hexane, metribuzin was then eluted with 5% acetone in hexane. Methylation of the diethyl ether extracts using diazomethane and the Florisil column cleanup of the methylated extracts were also as described previously (Cessna et al., 1985).
Gas Chromatographic Analysis
Quantification and confirmation of herbicide residues in the drainage water extracts were performed using a Hewlett-Packard (Palo Alto, CA) Model 5890A gas chromatograph (GC) interfaced to the Model 5970B mass selective detector (MSD), which was operated in the selected ion monitoring mode. The GC–MSD system was controlled with the Model 5895A data station and the GC was equipped with a 25-m by 0.2-mm-i.d. Ultra-1 capillary column (Hewlett-Packard; film thickness of 0.11 µm). Injections (2 µL) of sample extracts were performed using the Model 7673A autoinjector. The split-splitless injector was operated in the splitless mode and maintained at a temperature of 230°C. The helium carrier gas flow was 25 cm s–1, and the column temperature program consisted of an initial temperature of 70°C for 1 min followed by a temperature increase of 5°C min–1 to 270°C and hold for 1 min. The capillary interface between the GC and the MSD was maintained at 280°C throughout each run. The retention times for the neutral herbicides and the methyl esters of the acidic herbicides under these operating conditions are presented in Table 1.
Three or four ions, characteristic of each neutral herbicide and each methylated acidic herbicide, were monitored for confirmation purposes. These ions and the relative response of the ions used for confirmation are presented in Table 1.
The presence of an herbicide was considered to be confirmed if (i) all ions monitored were present, (ii) a peak appeared at the retention time (±0.02 min) obtained for a standard solution of the herbicide in the reconstructed chromatograms of all ions monitored, and (iii) the peak area ratio was within ±30% of the ratio obtained using a standard solution of the herbicide.
Recoveries of the herbicides from water fortified at 0.1 and 1.0 µg L–1 ranged from 63 to 107% with the exception of fenoxaprop for which recoveries ranged from 45 to 68%. The limit of quantification of the analytical method was 0.05 µg L–1 at a signal to baseline noise ratio of 3:1. Trace concentrations were considered to be >0.01 to <0.05 µg L–1.
Dugout Watershed Areas and Characterization
The Tisdale watershed was farmed as a single field whereas those near Lucky Lake and Craik were split into two fields of approximately equal size. The area of the watershed draining into each dugout was determined by conducting a topographic survey using a theodolite (Sokkia set5 laser theodolite; Sokkia Corporation-Canada, Mississauga, ON). The elevation data were used to delineate the boundaries of the basins. A snow survey was conducted on the Tisdale watershed before snowmelt in all three years to calculate the volume of water held in the snow pack. On the other watersheds, surveys were conducted only if there was significant snow pack at the beginning of March. This occurred at Lucky Lake in 1995 and Lucky Lake and Craik in 1997.
Dugout Volumes
Gauge plates (Water Survey of Canada standard gauge plates) were installed in each dugout and the depth determined to the nearest centimeter. The volume of water contained in a dugout was calculated from the dimensions of the dugout and the depth of the water in it. The Tisdale and Craik dugouts were constructed by the Prairie Farm Rehabilitation Administration (PFRA) and their dimensions were obtained from specifications provided by PFRA. The dimensions of the Lucky Lake dugout were measured using a theodolite in fall 1995 when water levels were low. Rainfall data for the vicinity of each dugout (<1 km from the Lucky Lake dugout to approximately 10 km from the Tisdale dugout) were obtained from Saskatchewan Agriculture and Food, Regina, SK and used to identify possible runoff events.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Water Quantity
In any year of the 3-yr study, water volume in the snow pack tended to decrease in the order Tisdale > Craik > Lucky Lake. By comparing the overwinter recharge to the dugouts to the volume of water present in snow in the catchments, it was estimated that less than 20% of the snow pack reached the dugouts in any year of the study. Overwinter recharge was generally greatest at the Craik site where snow fences were used to trap snow next to the dugout. The Tisdale dugout, which overflowed in 1995 and 1996, was filled to capacity in all three years. The Craik dugout was filled to capacity in 1996 and 1997, and the Lucky Lake dugout only in 1996. In spring 1995, there was relatively little snowmelt runoff at the latter two sites and water volumes immediately after snowmelt runoff were well below capacity in these dugouts.
Following snowmelt runoff, water volumes of all three dugouts generally decreased continuously due to evaporative and possible seepage losses, household and potable water use at the Craik and Tisdale sites, and livestock watering at the Lucky Lake site. During the 3-yr study, seasonal decreases in water volume ranged from 20 to 30% for the larger dugouts (Craik and Tisdale) and 80 to 90% for the smallest dugout and corresponded to decreases in water depth of approximately 0.6 and 1.6 m, respectively. As water depth in each dugout decreased, the distance between the intake of the sampling tube and the bottom of the dugout also decreased for the collection of the 1995 and 1996 samples. At the Lucky Lake site in 1995, water depth decreased to the extent that, beginning in July, only single samples at half depth were subsequently collected.
Individual rainfalls at the three sites varied from <5 to 57 mm. In 1996, when water depths were measured every 5 d, there was no evidence for the occurrence of surface runoff at the Tisdale and Craik sites. Following major rainfalls at the Craik site, water volume increased after several days suggesting that the increases may have been due to influxes of surficial ground water. At the Lucky Lake site, a very intense 57-mm rainfall in June produced significant surface runoff, which caused the dugout to overflow and flood an area of several hectares around it.
Water Quality
Herbicide Concentrations in the Dugout Waters
Winter Samples. Herbicides were detected in the water of all three dugouts: 2,4-D at Lucky Lake, 2,4-D and dicamba at Craik, and 2,4-D, clopyralid, and diclofop at Tisdale (Table 3). Of the herbicides applied during the 1994 growing season (Table 2), only those applied to the Craik watershed were detected in the winter samples.
|
Lucky Lake.
No herbicides were detected in the sample collected from on top of the ice during the brief period of snowmelt in mid-March 1996. When snowmelt resumed in early April, herbicides (dicamba, 2,4-D, bromoxynil, and MCPA) applied to the watershed during the 1995 growing season (Table 2) were occasionally detected in the runoff water whereas trifluralin, applied the previous fall, was not (Table 4). In 1997, when snowmelt runoff occurred over a relatively short time frame and only two samples were collected, none of the herbicides (MCPA, bromoxynil, ethalfluralin) applied to the watershed during the 1996 growing season were detected in the runoff water; however, trace concentrations of trifluralin, last applied in fall 1995, were present.
|
|
|
|
In 1996, three (2,4-D, dicamba, and MCPA) of the four herbicides detected at the termination of snowmelt runoff were still present in the dugout water when the first 5-d composite sample was collected in May (Table 8). There was a major rainfall (57 mm) and surface runoff event on 17 June that caused the dugout to fill and overflow. Ethalfluralin, soil-incorporated into the western portion of the watershed (Table 2), was not detected in the dugout water until after this runoff event when trifluralin, applied in fall 1995, was also detected. Postemergence applications of metribuzin and a tank mixture of bromoxynil plus MCPA on 22 June did not result in detectable concentrations of these herbicides in the dugout water as a consequence of application drift. When the water volume had decreased by approximately 70% (last week in August), ethalfluralin, MCPA, and bromoxynil were detected.
|
|
In 1996, diclofop was not present in the dugout water during runoff but was detected in relatively high concentrations (0.45–1.07 µg L–1) from the initial sampling in May until the end of July (Table 5). During this period, 2,4-D was detected almost continuously whereas MCPA, bromoxynil, and dicamba were detected infrequently and only in trace concentrations. Detectable concentrations of the latter three herbicides did not result from drift during the application of tank mixtures of 2,4-D plus dicamba and MCPA plus bromoxynil in June (Table 2).
Detectable concentrations of herbicides reoccurred in the dugout water with the 15 to 20 September sampling, at which time MCPA, bromoxynil, 2,4-D, dicamba, and diclofop were detected. With the exception of diclofop, concentrations were generally much greater than those during the June to August period (Table 5). Trifluralin and triallate subsequently appeared in the 20 to 25 September sample. With the exception of bromoxynil, the six herbicides were then detected continuously until the last sampling (15–20 October) when only 2,4-D, dicamba, and triallate were detected.
Of the seven herbicides detected late in fall 1996, only 2,4-D, dicamba, and trifluralin were detected in the first sample collected in spring 1997 (Table 5). These three herbicides were also the only herbicides present during snowmelt runoff (Table 4). Trifluralin was not detected in any subsequent samples. The presence of 2,4-D and dicamba in the dugout water in April probably accounts for their detection following application of a 2,4-D plus dicamba tank mixture in June, rather than inputs from application drift.
Tisdale.
Of the herbicides (MCPA and dicamba) applied to the watershed in 1994 (Table 2), only MCPA was detected in the dugout water in 1995 (Table 3). Although 2,4-D, clopyralid, and diclofop were not applied in 1994, they were present in the overflow water during snowmelt runoff (Table 6) and were detected in the dugout water (2,4-D and clopyralid continuously). Following spring application of the 2,4-D plus dichlorprop tank mixture in June, dichlorprop was not detected in the dugout water as result of drift deposition.
In 1996, 2,4-D, dicamba, and MCPA, which were present in the overflow water at the end of runoff (Table 6), were still detectable in the dugout water (Table 7). However, by mid-June, herbicide concentrations had decreased to nondetectable levels, even though a tank mixture of 2,4-D plus dichlorprop had been applied. With the exception of the last two weeks in August when 2,4-D, dichlorprop, bromoxynil, and MCPA were present, no herbicides were detected in the dugout water until, as with the Craik dugout, the latter part of September. At this time, detectable concentrations of several herbicides (2,4-D, dicamba, MCPA, and bromoxynil) reappeared in the dugout water that were much higher than those observed during June and July. Somewhat later, triallate, trifluralin, and dichlorprop were also present in the dugout water.
In 1997, 2,4-D, detected continuously during runoff (Table 6) and the previous fall (Table 7), was also continuously detected in the dugout water (Table 9). Detected during runoff, dicamba was subsequently detected in the dugout water whereas trifluralin was not. Bromoxynil, applied to the watershed in June 1997, was not detected in the dugout water
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Surface water bodies also receive pesticide inputs through the atmospheric processes of wet (precipitation) and dry atmospheric deposition. Wet and dry deposition of pesticides has been measured in Saskatchewan (Waite et al., 2002) and bulk (dry plus wet) deposition in Alberta (Hill et al., 2001) and Manitoba (Rawn et al., 1999). In a recent study, in which wetlands situated in wildlife habitat and on farms of high (zero tillage), moderate (conventional tillage), and no (organic farming) pesticide inputs were monitored for herbicide content, Donald et al. (2001) found that frequency of detection and concentrations of individual herbicides were similar regardless of land-use type. The authors suggested that atmospheric processes could account for both the concentrations and relatively uniform distribution of herbicides in the wetlands on all landscape types. These processes may similarly affect farm dugouts.
For several of the herbicides currently used in crop production systems on the Canadian prairies, entry into farm dugouts and other surface waters via surface runoff (Wauchope, 1978; Spencer and Cliath, 1991) or atmospheric deposition (Waite et al., 2002; Rawn et al., 1999; Hill et al., 2001) would have been ongoing for several decades. 2,4-D and MCPA have been in use in the prairie region for more than 50 yr, dicamba, mecoprop, triallate, and trifluralin for about 40 yr, bromoxynil for approximately 30 yr, diclofop for 20 yr, and clopyralid and ethalfluralin for about 10 yr (Ahrens, 1994). Consequently, the Tisdale dugout, constructed 22 yr before the end of the present study, may have received atmospheric deposition and surface runoff inputs of 2,4-D, MCPA, dicamba, mecoprop, triallate, and trifluralin since its construction whereas inputs of clopyralid and ethalfluralin could have occurred only during the last 10 to 12 yr. Since the corresponding ages of the Lucky Lake and Craik dugouts were 11 and 14 yr, respectively, both dugouts may have received inputs for all of these herbicides since their construction.
The presence of herbicides in water of all three dugouts during the winter (February) of 1995 (Table 3) when the dugouts were covered by ice was most likely a consequence of long-time seasonal input. Gao et al. (1998a) have showed that the sorptive capacity of pond sediment for several pesticides was directly related to its organic carbon content and particle size and that the adsorption process was not completely reversible. These researchers (Gao et al., 1998b) also showed that aged (3–7 yr) residues of atrazine and bifenox [methyl 5-(2,4-dichlorophenoxy)-2-nitrobenzoate] were desorbed from pond sediment in much smaller amounts than residues in freshly fortified sediment. Consequently, the herbicides detected in the dugout water in February 1995 may have been present in the bottom sediments that, under appropriate environmental conditions, acted as a source of the herbicides to the water column.
Herbicides in Dugout Waters
Of the 13 herbicides monitored in the dugout water (Table 1), three were soil-incorporated and included triallate (not applied to the dugout watersheds during the study), trifluralin, and ethalfluralin. These herbicides, applied either in the fall or early spring (May), have relatively low water solubilities and sorb more strongly to soil (Table 10). Due to their relatively high vapor pressures, the main route of dissipation of these herbicides from soils is volatility losses to the atmosphere (Grover et al., 1988; Glotfelty et al., 1984). Trifluralin and ethalfluralin are susceptible to photodegradation (Parochetti and Dec, 1978), either when present on the soil surface or in runoff water.
|
Winter Samples
Water samples were collected during winter only in 1995. The herbicides detected in these samples from the Lucky Lake (2,4-D) and Tisdale (2,4-D, clopyralid, and diclofop) dugouts (Table 3) had not been applied to their respective watersheds the previous growing season (Table 2). Their presence in the dugouts may reflect release from bottom sediments during ice cover. At Craik, the herbicides detected (2,4-D, dicamba) had been applied the previous growing season and, although bottom sediments may also account for these residues, both herbicides may have entered the dugout through atmospheric and/or surface runoff processes.
Snowmelt Runoff Samples
Snowmelt runoff water per se was not collected at any site during the study. At Lucky Lake and Craik in 1996 and 1997 and at Tisdale in 1997, dugout water samples were collected during snowmelt. At Tisdale in 1995 and 1996, overflow water was collected. Thus, in all situations, samples were collected after the snowmelt runoff water had mixed with the dugout water.
The 1995 spring runoff provided the only comparison of herbicide concentrations in winter with those detected in the dugouts during runoff. At Tisdale in 1995, concentrations of 2,4-D and clopyralid in the overflow water (Table 6) were decreased relative to winter concentrations in the dugout water (Table 3) suggesting dilution of herbicide residues in the dugout water by snowmelt runoff. Because none of the herbicides detected (2,4-D, clopyralid, diclofop) in the winter samples had been applied to the watershed during the previous growing season (Table 2), dilution by the runoff water would have been expected. Decreased herbicide concentrations in the May dugout samples from Craik compared with those in winter probably occurred for the same reason, even though the herbicides detected (2,4-D and dicamba) had been applied the previous spring to the western half of the watershed. Dilution at this site indicates that, due to relatively short field half-lives, high water solubilities, and low soil sorption (Table 10), there was little persistence of either herbicide in the soil–runoff interaction zone during the 1994 growing season. At Lucky Lake, 2,4-D concentrations in the dugout water were greater after snowmelt runoff suggesting that the herbicide loading in the runoff water was greater than that in the dugout. This, however, seems unlikely because 2,4-D was not applied to the Lucky Lake watershed in 1994. Although it is not clear why a concentration increase occurred at Lucky Lake, it may be possible that these herbicides, if originally sorbed to bottom sediments, may have partitioned into water adjacent to the sediments under ice cover and then been dispersed throughout the dugout by mixing during snowmelt runoff and ice melt.
In all snowmelt runoff events monitored, herbicides were detected in the resulting mixture of dugout and runoff water (Tables 4 and 6). At Tisdale, where daily samples were collected during snowmelt each year, herbicide concentrations in the overflow (1995 and 1996) and dugout (1997) water tended to increase with increasing days of runoff. A similar pattern was observed at Craik in 1997 and at Lucky Lake in 1996. It was not possible to discern a pattern at Lucky Lake in 1997 because, due to reduced snowmelt volume, only two samples were collected. If the detected herbicides had originated in the dugouts as discussed for the Lucky Lake runoff in 1995, this pattern may reflect increased mixing of dugout and snowmelt water as snowmelt runoff proceeded. Fall application would provide the most likely scenario for herbicides detected during snowmelt runoff to originate in the runoff. In the present study, the only fall application was that of trifluralin (Table 2) to the Lucky Lake watershed in 1995. However, trifluralin, which has low water solubility, strong soil sorption, and a high vapor pressure (Table 10), was not detected during snowmelt runoff.
Spring to Fall Samples
The frequency of detection of herbicides in the dugout waters tended to reflect the length of time that the herbicide had been used in the prairie region and the extent to which they were used. The herbicide 2,4-D was detected in all dugouts at some point during the growing season each year (Tables 3, 5, 7, 8, and 9), which probably reflects its extensive and continuous use for more than 50 yr. Other herbicides detected in decreasing order were MCPA > dicamba > trifluralin > bromoxynil. The lower frequency of detection of bromoxynil and trifluralin may reflect lower use as well as their greater susceptibility to sunlight photolysis (Cessna and Muir 1991) and, in the case of trifluralin (Table 10), higher volatility.
Generally, herbicides detected in dugout water samples taken in winter through the ice or during snowmelt runoff were also detected in the initial sample collected in April or May (Tables 3– 9). From levels detected in the April and May samples, concentrations then tended to increase until mid-August. This increase in concentration may reflect increasing water temperature and greater partitioning of the herbicides from sediments into the water column, as well as inputs from application drift and wet and dry atmospheric deposition. By June and July 1995, there was evidence of concentration differences at shallow and deep depths (Table 3) indicating that some thermal stratification of the water bodies may have occurred. As a consequence, only the warmer upper layer would have been mixed by wind action on the surface of each dugout (Mackay, 1999a).
Evidence for significant thermal stratification of dugouts was reported by Mackay (1999a)(1999b) in a 2-yr study (1997–1998) in which various methods of aerating dugouts were evaluated. Dissolved oxygen concentration and water temperature were measured with depth at 1-m intervals in 14 non-aerated dugouts every two weeks from May through to October. In these dugouts, gradients in both dissolved oxygen concentration and temperature were evident throughout the sampling period. All but three dugouts were anoxic (<1 mg L–1 dissolved oxygen) at 2-m (n = 4) and 3-m (n = 7) depths suggesting that mixing of the water due to wind action on the dugout surface was to a depth of approximately 2 m. This was accompanied by a gradient of decreasing temperature with depth in all dugouts. The steepness of this gradient from the water surface to a 3-m depth, which varied greatly depending on the dugout and sampling time, ranged from 1 to 15°C. Thus, at times during the measurement period, some of the dugouts approached significant thermal stratification. In mid- to late September, when surface temperatures become similar to bottom temperatures and wind action mixes the entire dugout (Mackay, 1999a), both the dissolved oxygen concentration and temperature gradients disappeared due to this mixing process, which is referred to as "fall turnover."
Application drift during the spring spraying season, in general, did not result in detectable herbicide concentrations in the dugout water or concentration increases. In 1995, the application of the 2,4-D plus dicamba and MCPA plus bromoxynil tank mixtures to the Lucky Lake watershed (Table 2) did not result in detectable concentrations of dicamba, MCPA, or bromoxynil (Table 3). Similarly, MCPA and bromoxynil were not detected following application of the tank mixture at Craik nor dichlorprop detected following application of the tank mixture with 2,4-D at Tisdale. During the 1996 growing season, similar results were observed with the tank mixtures of bromoxynil plus MCPA and 2,4-D plus dichlorprop applied to the Lucky Lake and Tisdale watersheds, respectively (Tables 7 and 8). In 1997, application of the tank mixture of bromoxynil plus MCPA to all three watersheds did not result in detectable concentrations of either herbicide in the June samples (Table 9). The fall application of trifluralin to one-half of the Lucky Lake watershed in 1995 did not result in detectable concentrations immediately after snowmelt runoff the following spring (Table 8); however, the herbicide was detected as a trace concentration in late October immediately following its application and may reflect deposition of wind-eroded sediment (and associated trifluralin) into the dugout (Larney et al., 1999).
From mid-August, herbicide concentrations then generally tended to decrease through to September, frequently to nondetectable levels (Tables 3 and 5–9). This decrease in concentration may reflect both photochemical (Cessna and Muir, 1991) and microbial degradation, as well as possibly increased partitioning into bottom sediments with decreasing water temperature.
In mid- to late September, dugout turnover was evident from the October 1995 samples (Table 3) when concentration differences between the shallow and deep samples had essentially disappeared. In 1996, the 5-d composite samples indicated that marked increases in herbicide concentrations occurred during fall turnover, which may be explained as follows. During the 4-mo period of thermal gradient conditions, herbicides present in bottom sediments may partition in the colder, lower undisturbed layer. Then, when the mixing associated with fall turnover occurs, the herbicides, originally in the lower layer, are dispersed throughout the dugout. This mixing would account for the sudden appearance of herbicides in the water column in mid- to late September and in concentrations much greater than those observed during the spring spraying season (Tables 5, 7, and 8). This explanation assumes that the minimal aeration of the Craik dugout (2 h per day) and aeration at a 1-m depth at Tisdale did not prevent the formation of a thermal gradient in these dugouts and that the intakes of the autosamplers at a 1-m depth were positioned in all three dugouts within the warmer, and relatively well-mixed, upper layer.
Some herbicides were detected in dugout water even though they had not been applied to the corresponding watersheds either during the previous or current growing seasons. For example, diclofop was detected in both the Craik and Tisdale dugouts in 1995 (Table 3) and in the Craik dugout in 1996 (Table 5) in relatively large concentrations. Clopyralid was similarly detected in the Tisdale dugout in 1995. Following fall turnover in 1996, diclofop and triallate were detected in the Craik dugout and bromoxynil, trifluralin, and triallate were detected in the Tisdale dugout (Table 7). These data, especially those associated from fall turnover, suggest that these herbicides may have been sorbed to bottom sediments and released to the water column. The data also suggest that these herbicides may have been resident in the sediments for periods up to 3 yr.
Grover et al. (1997), in their study of the magnitude and persistence of herbicides in 21 farm dugouts, suggested that median concentrations, which were generally near or below 0.05 µg L–1, may be an indicator of the general contamination of farm dugouts within the prairie region due, in part, to long-term seasonal inputs through snowmelt and rainfall runoff, application drift, and atmospheric deposition. In the present study, using monthly samples from 1995 and 1997 and mid-monthly 5-d composite samples from 1996, median concentrations of all herbicides, with the exception of 2,4-D, were less than 0.05 µg L–1.
Maximum herbicide concentrations did not exceed Canadian drinking water guidelines (Table 11) but, of the 181 water samples analyzed in the current study, herbicide concentrations in 85 samples (47%) exceeded the more stringent guidelines established by countries of the European Union for drinking water: 0.03 to 0.10 µg L–1 for individual pesticides and 0.50 µg L–1 for mixtures of pesticides (Council of the European Union, 1998). Herbicide concentrations exceeded Canadian water quality guidelines for aquatic life in approximately 11% of the samples although guidelines for clopyralid, which was detected in 23 samples, have not yet been established. Irrigation water guidelines were exceeded in 40% of the samples and occasionally for more than one herbicide in a single sample. This is probably somewhat of an underestimation for two reasons: the irrigation water guideline for dicamba (0.006 µg L–1) would have been considered a trace concentration in the present study and guidelines for clopyralid, trifluralin, and triallate have not yet been established. Samples in which herbicide concentrations exceeded all three guidelines were typical of those collected following fall turnover of the dugouts.
|
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. Farenhorst, S. K. Papiernik, I. Saiyed, P. Messing, K. D. Stephens, J. A. Schumacher, D. A. Lobb, S. Li, M. J. Lindstrom, and T. E. Schumacher Herbicide Sorption Coefficients in Relation to Soil Properties and Terrain Attributes on a Cultivated Prairie J. Environ. Qual., May 1, 2008; 37(3): 1201 - 1208. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. T. Waite, A. J. Cessna, R. Grover, L. A. Kerr, and A. D. Snihura Environmental Concentrations of Agricultural Herbicides in Saskatchewan, Canada: Bromoxynil, Dicamba, Diclofop, MCPA, and Trifluralin J. Environ. Qual., September 1, 2004; 33(5): 1616 - 1628. [Abstract] [Full Text] [PDF]
|
|
