William A. Battaglin, Hydrologist

USGS, Box 25046, MS 406, Denver, CO 80225, Phone: (303) 236-5950 x202

Edward T. Furlong, Research Chemist

USGS, Box 25046, MS 407, Denver, CO 80225, Phone: (303) 467-8080

Mark Burkhardt, Research Chemist

USGS, Box 25046, MS 407, Denver, CO 80225, Phone: (303) 467-8093

C. John Peter, Research Associate

DuPont Agricultural Products, Barleymill Plaza, PO Box 80015, Wilmington, DE 19880-0015

Phone: (302) 992-2277

A new generation of low-application rate herbicides that function by inhibiting the action of a key plant enzyme are gaining popularity among farmers. Sulfonylurea (SU), sulfonamide (SA), and imidazolinone (IMI) herbicides are classes of herbicides that function in this manner. These herbicides are applied either pre- or post-emergence to crops at rates ranging from 0.001 to 0.25 pound active ingredient per acre. This is much less than the application rate of many other corn and soybean herbicides. SUs, SAs, and IMIs have very low toxicities to mammals and other animals, but like other herbicides they can cause problems with nontarget plants even when only small amounts of the originally applied material remain in the soil or in water. Little is known about the occurrence, fate, or transport of these new herbicides in surface water or ground water in the United States. The purpose of this reconnaissance is to gain additional information and understanding about the occurrence of selected SUs, SAs, and IMIs in water resources of the midwestern United States, the area of highest use for these compounds. This study will build upon knowledge and information gained in earlier herbicide reconnaissance studies conducted by the U.S. Geological Survey (USGS) between 1989 and 1997. Approximately 200 samples from small streams, larger rivers, reservoir outflows, and wells will be collected and analyzed in 1998. The study is the result of a cooperative research and development agreement between the USGS and E.I. DuPont.

During the last 20 years, a generation of low application rate herbicides has been developed that act by inhibiting the action of a key plant enzyme, resulting in stopped growth and eventual plant death. These herbicides are gaining in popularity among farmers. Sulfonylurea (SU), sulfonamide (SA), and imidazolinone (IMI) herbicides are three classes of compounds that share the same mode of action. These compounds typically are applied at much lower rates than triazine or acetanilide herbicides. Crops that can be treated with SUs, SAs, and IMIs include barley, corn, cotton, durum wheat, rice, canola, peanuts, soybeans, sugar beets, spring wheat, and winter wheat. Some of these compounds are also approved for use on Conservation Reserve Program acreage (land set aside from crop production generally because farming it could result in significant soil loss) and for noncropland weed control. The total corn, soybean, and wheat acreage on which 9 SUs, 1 SA and 2 IMIs were applied in eleven midwestern States (Iowa, Illinois, Indiana, Kansas, Kentucky, Minnesota, Missouri, Nebraska, Ohio, South Dakota, and Wisconsin) from 1990 through 1997 is shown in figure 1 (U.S. Department of Agriculture, 1991-98). In 1997, the area treated exceeded 66 million acres. For comparison, atrazine, a triazine herbicide, was used on about 42 million acres of corn in the same 11-State area in 1997.

Although applied over comparable areas, SU, SA, and IMI herbicides typically are applied at postemergence at extremely low rates (0.001 to 0.25 pounds of active ingredient per acre) These application rates are much lower than for other commonly used herbicide classes; hence, their overall use amount is small. For example, in 1996, in the same 11-State midwestern area, an estimated 23,200 tons of atrazine and 19,360 tons of metolachlor were applied to cropland, while the estimated use of 9 SUs, 1 SA, and 2 IMIs was only 1,150 tons (U.S. Department of Agriculture, 1997).

The soil half-life of SUs, SAs, and IMIs generally ranges from 1 to 25 weeks depending on soil pH and temperature. The water solubility of SUs, SAs, and IMIs generally ranges from 6 to 40,000 part per million. The water solubility of SUs is dependent on water pH, and the water solubility of SAs and IMIs is dependent on temperature and soil moisture content.

SUs, SAs, and IMIs act upon a specific plant enzyme (acetolactate synthase) that is not found in mammals or other animals and they are reported to have very low toxicities in animals (Brown, 1990, Meister, 1997). Plants demonstrate a wide range in sensitivity to SUs, SAs, and IMIs (Peterson et al., 1994) with over a 10,000 fold difference in observed toxicity levels for some compounds. The EC50 (concentration causing a 50 percent reduction in a chosen plant characteristic for which a toxicity endpoint exists, for example lab tests measuring biomass development) values for 5 aquatic plants are shown on figure 2 (U.S. EPA, 1997; Sabater and Carrasco, 1997; Fairchild et al., 1997). The EC50 values plotted are for green algae (Selenastrum capricornutum), duckweed (Lemna gibba), blue-green algae (Anabaena flos-aquae), freshwater algae (Scenedesmus costatum), and freshwater diatom (Navicula pelliculosa). In some cases, EC50 values from more than one test on the same plant species are included. EC50 values for several herbicides range over 3 orders of magnitude. The EC50 data plotted on figure 2 support the hypothesis that a concentration of 0.1 mg/L in water is the baseline for non-target aquatic plant toxicity.

Crop toxicities reported as EC25 values (application rates in pounds per acre resulting in a 25% reduction in a chosen plant characteristic for which a toxicity endpoint exists, for example lab tests of juvenile plant growth) were used to estimate the herbicide concentrations in soil water that could result in non-target crop stress (U.S. EPA, 1997). EC25 values ranged from less than 0.00002 to more than 1.0 pound per acre. The soil water concentrations resulting from the EC25 application rate were calculated assuming that the upper portion of an acre of a typical cropped loam soil has approximately 50 percent pore space which under ideal growing conditions is occupied half by water and half by air (Buckman and Brady, 1969). Thus, the top 1 foot of soil would contain 3 inches of water and this water would weigh 679,100 pounds. If we assume that rain/irrigation water distributes the mass of herbicide equal to the EC25 value to this water and no herbicide degradation, then the herbicide concentration in that soil water can be estimated as:

Soil water concentration in parts per billion (ppb) = (EC25 (lbs/acre) / (679,100 (lbs/acre)) * 10⁹ (1)

For example, using equation (1) the concentration of nicosulfuron in soil water which may be of concern for juvenile onion plants can be estimated as:

concentration in ppb= (0.0039 (lbs/acre)/ (679,100 (lbs/acre)) * 10⁹ = 5.7 ppb.

Estimated soil water concentrations computed using the above equation for 9 crops are shown on figure 3. Note that 1 ppb is equal to 1 mg/L. These estimates do not account for herbicide degradation. The crops for which data are plotted are corn, soybeans, sorghum, canola, sugar beets, radish, onion, lettuce, and cabbage. Data on figure 3 indicate that non-target crop stress is unlikely to occur when soil water concentrations of SUs, SAs, and IMIs remain below 0.01 mg/L. Herbicide performance and non-target crop toxicity have been reported to vary by soil pH, soil organic content, and climate (Blair and Martin, 1988).

Because SUs, SAs, and IMIs are active at very low concentrations, they can cause a problem with plant vigor in some crop rotations even when only 1 percent or less of the originally applied material remains. Some of these herbicides have demonstrated residual phytotoxicity to rotation crops like corn, sunflowers, sugar beets, and dry beans for a year or more after application (Anderson and Barrett, 1985; Anderson and Humburg, 1987). Marrs et al. (1989) noted that a buffer of 5-10 meters is needed for ground sprayers to minimize SU impacts on non-target native plants. Fletcher et al. (1993) indicated that spray drift containing SUs at concentrations less than 1 percent of the recommended application rate may adversely impact fruit tree yields. Felsot et al. (1996) suggested that the appearance of chlorotic spots on crops in south central Washington is a result of exposure to low levels of SU herbicides in precipitation and not from direct spray drift. However, Obrigawitch et al. (1998) question the validity of Fletcher’s findings and the results of other studies that based their findings on short-term plant-response assessments. In an extensive review of existing field data, Obrigawitch et al. (1998) found that a treatment rate of 0.1 gram of the most active SU ingredient per hectare (0.00009 pound per acre) represents a "threshold dose" and would be unlikely to reduce the yields of even the most sensitive non-target plants. This threshold dose represents 1 percent or less of the use rates of SUs. Using the equation above, this dose would result in a soil water concentration of 0.13 mg/L. The data shown on figure 3 support Obrigawitch’s "threshold dose" hypothesis. The data shown on figures 2 and 3 suggest that there is a potential for SU, SA, and IMI herbicide to have an adverse effect on aquatic plants or crops if they were to occur in river water, rainwater, irrigation water, or soil water at concentrations greater than 0.1 mg/L.

As of July 1998, detections of SUs, SAs, and IMIs in water collected from environmental settings have been rare and the few reported detections have been at nanogram per liter concentrations (Bergstrom, 1990; Michael and Neary, 1993; D’Ascenzo et al., 1998; Steinheimer et al., 1998). However, several studies indicate that some SUs, SAs, and IMIs herbicides may leach beyond the active root zone and enter ground water or surface water systems (Anderson and Humburg, 1987; Bergstrom, 1990; Flury et al., 1995; Veeh et al., 1994). Once in ground water or surface water, some SUs, SAs, and IMIs would tend to persist as the parent compound while others would tend to hydrolyze (Dinelli et al., 1997; Harvey et al., 1985). A study by Afyuni et al. (1997) indicated that between 1.1 and 2.3 percent of an applied SU was lost in runoff during a simulated rainfall event 24 hours after herbicide application.

Because of their low application rates and low overall use amounts, concentrations of SUs, SAs, and IMIs are expected to be low or nondetectible in most water resources. Environment Canada uses a calculated Expected Environmental Concentration (ECC) to evaluate the potential hazard of pesticides to nontarget aquatic organisms. ECC values for selected SUs ranged from 3 to 20 mg/L and equaled 67 mg/L for imazethapyr (Peterson et al., 1994). For comparison, ECC values for selected triazine herbicides were more than 100 times greater at 2,667 to 2,867 mg/L. These are worst case exposure estimates and assume overspray of a 15 cm deep water body with the maximum application rate.

One can also assume based upon their chemical characteristics, application rates, and acres treated that individual SUs, SAs, and IMIs herbicides would be expected to occur in surface or ground water at 1 to 0.1 percent or less of the concentration of common triazine herbicides. The USGS measured concentrations of 11 common herbicides and 2 herbicide metabolites in samples from 52 Midwestern rivers during runoff events that occurred soon after herbicide application in 1989, 1990, 1994 and 1995 (Goolsby et al., 1994; Battaglin and Goolsby, 1996). The median concentrations in these samples represent an estimate of expected environmental concentrations based upon observation. Median atrazine concentration for the four years of data ranged from 5.5 to 10.9 mg/L; median cyanazine concentrations ranged from 1.3 to 2.7 mg/L; and median metolachlor concentrations ranged from 1.7 to 2.5 mg/L. Maximum concentrations for these three compounds for the 4 years ranged from 10.6 to 108 mg/L. Thus, one could expect to observe SUs, SAs, and IMIs herbicides in Midwestern rivers during post-application runoff events at concentrations ranging from 0.001 to 0.1 mg/L. Further, one could expect maximum concentrations of SUs, SAs, and IMIs herbicides to range from 0.01 to 1.0 mg/L.

Currently, little is known about the occurrence, fate, or transport of SUs, SAs, and IMIs in surface water and ground water in the United States. The overall objective of this project is to determine if and at what concentrations selected SUs, SAs, and IMIs occur in surface and ground water resources of the midwestern United States. Specific objectives include:

Specific hypotheses to be tested are:

5. The frequency of detections and concentrations of SU, SA, and IMI herbicides will be greater in smaller watersheds that are predominantly agricultural than in larger watersheds that have more diverse land use and land cover.

This study will involve collecting approximately 200 samples during a 1998 reconnaissance. Samples will be collected from streams, large rivers, reservoir outflows, and wells. When and where possible samples will be collected in conjunction with USGS National Stream Quality Accounting Network (NASQAN) and National Water Quality Assessment (NAWQA) activities, both to reduce the cost of sample collection and to insure availability of QA/QC and other water-quality data (other herbicides, insecticides, and nutrients). All 1998 herbicide reconnaissance samples will be sent to the Methods Research and Development Program personnel at the USGS National Water Quality Lab (NWQL) in Arvada, Colorado, and analyzed for a minimum of 16 herbicides (table 1) using high performance liquid chromatography coupled with mass spectrometry. This analytical method will have a quantification limit of less than 0.1 mg/L for all analytes. Samples will also be analyzed for other pesticides and nutrients by standard methods.

One hundred sites will be sampled in this study (figure 4). Samples will be collected from 75 surface-water sites in the Upper Mississippi, Missouri, and Ohio River basins. Fifty-two of the surface water sites to be sampled have been studied in previous Midcontinent Herbicide Initiative (MHI) investigations (Thurman et al., 1992; Goolsby et al., 1994). These sites were selected out of the set of 150 sites sampled in 1989 using a stratified random method (Scribner et al., 1993). The original 150 surface water sites were selected from the population of all USGS gaging stations in a 10-State area using a stratified random-sampling procedure to ensure adequate geographic distribution. The number of sites per State was proportional to corn and soybeans production, and sites were randomly selected by county within the State. The sites sampled in 1989 were ranked according to the total herbicide concentration in the post-application sample. These sites were then divided into three equal groups and 26 sites were randomly selected from the highest concentration group while 13 sites were randomly selected from each of the other two groups. It is important to note that this sampling strategy is not designed to produce an unbiased estimate of herbicide occurrence in all midwestern streams. Rather the intent was to target higher risk areas while still capturing the variability of the entire population. Another reason for sampling the 52 sites is the considerable historical data at these sites which will enable us to put in context the concentrations measured in 1998 samples with those measured in 1989, 1990, 1994 and 1995.

Samples also will be collected at selected NASQAN and NAWQA sites and just downstream from five reservoirs at locations that were sampled in a previous investigation (Coupe et al., 1995; Scribner et al., 1996). The majority of ground water samples will be collected from a network of wells in Iowa that are part of the Iowa Ground Water Monitoring (IGWM) program (Detroy et al., 1988; Kolpin et al., 1997). Wells from this network have been sampled systematically since 1982. Samples will also be collected from selected wells in the Lower Illinois NAWQA study unit.

Two samples will be collected at each surface-water and reservoir site (one on each of two site visits), and one sample will be collected at each ground water site. The first surface-water samples will be collected after pre-emergence herbicides have been applied (usually May or June) and following a precipitation event that produces a significant increase in streamflow. Ideally, streamflow should be representative of runoff conditions with flow at or above the 50th percentile (50 percent exceeds values for the period of record, published in State USGS Water Resources Data reports). These samples will be referred to as pre-emergence runoff samples. The second surface-water samples will be collected after post-emergence herbicides have been applied (usually June or July) again following a precipitation event that produces runoff conditions and streamflows at or above the 50th percentile. These samples will be referred to as post-emergence runoff samples. The first NASQAN and reservoir samples will be collected in May or June, 2-3 weeks after the first surface-water samples were collected from nearby sites. The second NASQAN and reservoir samples will be collected in June or July, 2-3 weeks after the second surface-water samples were collected from nearby sites. Collection of these stream and reservoir samples may require special visits to sites. Ground-water samples will be collected in June, July or August.

Samples will be collected using protocols that are identical to those used for the collection of samples for low levels of other dissolved organic compounds (Shelton, 1994), such as the NWQL schedule 2001 which is a capillary-column gas chromatography/mass spectrometry method for pesticide analysis, or the schedule 2050 which is a high-pressure liquid-chromatography method for pesticide analysis. The equal-width-increment sampling method (Edwards and Glysson, 1988) is recommended, but equal-discharge-increment sampling is also acceptable on larger rivers. All equipment will be precleaned with a Liquinox/tap-water solution, rinsed with tap water, deionized water, and then methanol, and air dried. All pesticide samples will be filtered through 0.7-mm pore-size baked glass-fiber filters using an aluminum plate filter holder and a ceramic or stainless steel piston fluid metering pump with all teflon tubing into precleaned 1-liter or 125-ml amber glass bottles. Samples will be immediately chilled and shipped on ice from the field to the labs within two days of collection.

Five 125-ml amber glass bottles from each site will be sent to the USGS laboratory in Lawrence, Kansas for analysis of herbicide compounds. Two 1-liter glass bottles from each site will be sent to the USGS NWQL in Denver for SU, SA, and IMI herbicide analysis. One 1-liter glass bottle and one 125-ml polyethylene bottle will be sent to the USGS NWQL in Arvada for pesticide (schedule 2001) and nutrient (schedule 2752) analysis. Field measurements for specific conductance, pH, and temperature will be taken for all samples and stream discharge will be obtained either by direct measurement, from a rating curve, or estimated from a nearby gaging station.

Methods Research and Development Program personnel at the USGS NWQL will validate and improve a routine analytical method, provided by DuPont Agricultural Products, for measuring trace (ng/L) concentrations of selected SU, SA, and IMI herbicides in water samples. The method will use high performance liquid chromatography (HPLC) coupled with mass spectrometry. The analytical method to be validated was developed by DuPont as one of five methods developed in association with the EPA/Industry Multianalyte Methods group. The DuPont method (Rodriguez and Orescan, 1996) uses electrospray liquid-chromatography/mass spectrometry (LC/MS) to detect 16 (table 1) SUs, SAs, and IMIs with a reporting limit of 0.1 mg/L for all analytes. Other analytical methods with similar or lower reporting limits for selected compounds have also been reported (Di Corcia et al., 1997; Nilve et al., 1994). Improvements to the DuPont method will include (1) switching from external standard quantitation to internal standard quantitation, (2) increasing the sample size for extraction from 250 ml to 1 liter, (3) testing several new extraction media, and (4) expanding the list of target analytes to include other SU, SA, and IMI herbicides and herbicide metabolites.

In addition, all samples will also be analyzed for several other classes of pesticides and for nutrients. Samples will be analyzed for 12 herbicides and 5 or more herbicide metabolites by gas chromatography/
mass spectrometry (GC/MS) using methods described by Thurman et al. (1990), Meyer et al. (1993) and Aga et al., (1994) by staff at the USGS lab in Lawrence. This method has an analytical reporting limit of 0.05 mg/L for most analytes. Samples will be analyzed for 41 pesticides and pesticide metabolites by GC/MS with selected-ion monitoring using methods described by Zaugg et al. (1995) by staff at the USGS NWQL. This method has analytical reporting limits that range from 0.001 to 0.018 mg/L. All samples will be analyzed for dissolved nitrite, nitrate plus nitrite, ammonia, and orthophosphate by an automated colorimetric procedure (Fishman and Friedman, 1989) by staff at the USGS NWQL. About 100 samples will be analyzed using both the original DuPont method and the modified DuPont method. Splits from these samples will also be sent out for confirmatory analysis at a DuPont laboratory.

Results will be presented that summarize the occurrence and distribution of SUs, SAs, and IMIs in midwestern water resources, and place their occurrence and distribution in relation to those of the other herbicides and herbicide metabolites measured in this study. If possible, estimates of the use of SUs, SAs, and IMIs or surrogates for those use estimates such as estimates of cropped land, will be used in statistical models that may help identify watershed conditions that influence SU, SA, and IMI concentrations.

Quality control (QC) samples will be collected at selected sites to provide information on the variability and bias of the measured SU, SA, and IMI concentrations. These samples will consist of concurrent replicates (CR), which are two samples collected as closely as possible in time and space, but processed, handled, and analyzed separately; laboratory spikes (LS), which are collected like concurrent replicates, then spiked in the lab with a known quantity of selected target analytes, and analyzed separately; or field blanks (FB), which are blank solutions that are subject to the same aspects of sample collection, field processing, preservation, transportation, and laboratory handling as the environmental samples. Concurrent replicates will be submitted blindly to the USGS NWQL, while laboratory spikes and field blanks will be identified as such. QC samples will only be collected and processed for the SU, SA, and IMI analysis. A total of 26 QC samples will be collected.

Water quality data will be managed using SAS software. Data will also be made available via the internet after it has been quality assured. Spatial data such as the location of sampling sites, the extent of drainage basins, and information about land use within those basins, will be managed using a geographic information system.

Where possible, sampling has been planned in conjunction with on-going NAWQA and NASQAN investigations to minimize costs and maximize information. Local USGS offices will be compensated for expenses associated with collection, processing, and shipment of samples to appropriate laboratories. The majority of costs associated with this effort will be covered by funds originating from a Cooperative Research and Development Agreement (CRADA) between the USGS and DuPont Agricultural Products (Battaglin et al., 1998).