Water, Air, & Soil Pollution

, Volume 218, Issue 1, pp 307–322

Occurrence of Azoxystrobin, Propiconazole, and Selected Other Fungicides in US Streams, 2005–2006

Authors

    • US Geological Survey
  • Mark W. Sandstrom
    • US Geological Survey
  • Kathryn M. Kuivila
    • US Geological Survey
  • Dana W. Kolpin
    • US Geological Survey
  • Michael T. Meyer
    • US Geological Survey
Article

DOI: 10.1007/s11270-010-0643-2

Cite this article as:
Battaglin, W.A., Sandstrom, M.W., Kuivila, K.M. et al. Water Air Soil Pollut (2011) 218: 307. doi:10.1007/s11270-010-0643-2

Abstract

Fungicides are used to prevent foliar diseases on a wide range of vegetable, field, fruit, and ornamental crops. They are generally more effective as protective rather than curative treatments, and hence tend to be applied before infections take place. Less than 1% of US soybeans were treated with a fungicide in 2002 but by 2006, 4% were treated. Like other pesticides, fungicides can move-off of fields after application and subsequently contaminate surface water, groundwater, and associated sediments. Due to the constant pressure from fungal diseases such as the recent Asian soybean rust outbreak, and the always-present desire to increase crop yields, there is the potential for a significant increase in the amount of fungicides used on US farms. Increased fungicide use could lead to increased environmental concentrations of these compounds. This study documents the occurrence of fungicides in select US streams soon after the first documentation of soybean rust in the US and prior to the corresponding increase in fungicide use to treat this problem. Water samples were collected from 29 streams in 13 states in 2005 and/or 2006, and analyzed for 12 target fungicides. Nine of the 12 fungicides were detected in at least one stream sample and at least one fungicide was detected in 20 of 29 streams. At least one fungicide was detected in 56% of the 103 samples, as many as five fungicides were detected in an individual sample, and mixtures of fungicides were common. Azoxystrobin was detected most frequently (45% of 103 samples) followed by metalaxyl (27%), propiconazole (17%), myclobutanil (9%), and tebuconazole (6%). Fungicide detections ranged from 0.002 to 1.15 μg/L. There was indication of a seasonal pattern to fungicide occurrence, with detections more common and concentrations higher in late summer and early fall than in spring. At a few sites, fungicides were detected in all samples collected suggesting the potential for season-long occurrence in some streams. Fungicide occurrence appears to be related to fungicide use in the associated drainage basins; however, current use information is generally lacking and more detailed occurrence data are needed to accurately quantify such a relation. Maximum concentrations of fungicides were typically one or more orders of magnitude less than current toxicity estimates for freshwater aquatic organisms or humans; however, gaps in current toxicological understandings of the effects of fungicides in the environment limit these interpretations.

Keywords

FungicideSoybean rustToxicityWater quality

1 Introduction

1.1 Problem

Fungicides are used in the US to prevent foliar diseases on a wide range of crops, and like other pesticides, can be transported from the fields where they are applied and contaminate the environment. Recent studies have detected fungicides and their primary transformation products (TPs) in streams, precipitation, groundwater, and bed sediment (McConnell et al. 1998; Wauchope et al. 2004; Scribner et al. 2006). Fungicide use is increasing in agricultural and nonagricultural sectors but the potential for a rapid increase in fungicide use due to soybean rust may be one of the most significant current causes. In some areas of the US, this may represent the first time that fungicides have been included in crop management practices.

Soybean rust is a devastating disease caused by two fungal pathogens: Phakopsora pachyrhizi (Asian species) and Phakopsora meibomiae (New World species). Asian soybean rust (soybean rust) was detected for the first time in the US in November 2004, in Louisiana. The concern for its arrival, however, preceded that date (Livingston et al. 2004). Subsequently, soybean rust has spread to an ever wider geographic area across the US (Sconyers et al. 2006; US Government Accountability Office 2006; US Department of Agriculture 2007). In 2005, detections of soybean rust were made in nine states and 131 counties (Fig. 1); in 2006, 15 states and 231 counties reported detections; and in 2007, 19 states and 335 counties (some as far north as Iowa), one province of Canada, and two municipalities in Mexico (US Department of Agriculture 2007).
https://static-content.springer.com/image/art%3A10.1007%2Fs11270-010-0643-2/MediaObjects/11270_2010_643_Fig1_HTML.gif
Fig. 1

Location of study sites and counties in the US with detected soybean rust in 2005 and 2006

The extent to which soybean rust will affect future US soybean production is unknown. The US Department of Agriculture has developed a plan to minimize the impact of soybean rust and to actively monitor for this disease (US Department of Agriculture 2004, 2007). Annual yield losses of 50% or greater in the southern US and a least 10% in the upper Midwest have been predicted (Dorrance et al. 2005).

1.2 Fungicide use

Fungicide use on crops in the US was estimated to be 52,000 metric tons per year in 2002 (the most recent year with estimates of US fungicide use; Gianessi and Reigner 2006). Some fungicides such as chlorothalonil have been in use for decades (first registered in 1966), whereas others such as azoxystrobin were recently introduced (first sales in 1996). Fungicides are used in the US to prevent foliar diseases on a wide range of crops (Meisterpro 2007). Typically, fungicides are more effective when used as protective rather than curative treatments. The primary crops on which fungicides are applied in terms of mass of active ingredient (circa 2002) are grapes (20,360 metric tons per year (mt/y)), tomatoes (4,680 mt/y), apples (3,310 mt/y), potatoes (3,270 mt/y), and citrus (3,180 mt/y). Other crops for which more than 50% of the planted acreage is generally treated with a fungicide include artichokes, brussel sprouts, cabbage, cucumbers, melons, peaches, peanuts, pumpkins, and sugarbeets (Gianessi and Reigner 2006). Fungicides are also used as biocides (i.e., wood preservatives); as antifungal agents for humans, pets, or livestock; and as a treatment for breast tumors (Zarn et al. 2003; Kahle et al. 2008).

The use of fungicides on corn and soybeans was minor prior to 2002. In 2002, less than 1% of US soybean acres were treated with a fungicide (US Department of Agriculture 2009a). In 2006, this use was already starting to increase, with 4% of soybeans receiving a fungicide application, most commonly pyraclostrobin or azoxystrobin (US Department of Agriculture 2009a). Fungicide use on corn was so minor that it was not reported by the US Department of Agriculture (2009a) in 2002 or in 2005 (the most recent year with a survey of agricultural chemical use on corn).

Another recent trend in crop management practices is to use fungicide treatments to increase yields of soybean and corn crops. Results of field trials suggest that “preventative” fungicide use can increase soybean yields by as much as 20% (Bradley and Sweets 2008; Yang et al. 2008), and corn yields by 5% or more (Syngenta FarmAssist 2009). Other researchers, however, suggests more evidence is needed before such claims can be substantiated (Ortiz-Ribbing et al. 2008; US Environmental Protection Agency 2009b).

In 2006, four fungicides were registered for treatment of soybean rust in the US: azoxystrobin, boscalid, chlorothalonil, and pyraclostrobin (US Environmental Protection Agency 2007), and five other fungicides were granted emergency exemptions for treatment of soybean rust: myclobutanil, propiconazole, tebuconazole, tetraconazole, trifloxystrobin. By 2009, 14 fungicides were registered for treatment of soybean rust; the nine previously mentioned plus cyproconazole, flusilazole, flutriafol, metconazole, and prothioconazole (West Virginia Department of Agriculture 2009). For the 12 target fungicide for this study (azoxystrobin, boscalid, chlorothalonil, cyproconazole, metalaxyl, metconazole, myclobutanil, propiconazole, pyraclostrobin, tebuconazole, tetraconazole, and trifloxystrobin) use increased about 25% from 4,210 metric tons in 2002–5,520 metric tons in 2007 (AgroTrak® agrochemical usage data (AgroTrak® is a registered trademark of dmrkynetec, St. Louis, MO, USA)). All 12 target fungicides are also registered for use on other crops or turf.

Therefore, there are many characteristics of fungicides that result in increasing usage and shifting spatial/temporal patterns in the US including: fungal diseases are more common than any other plant diseases; uses transcend agricultural sectors and include domestic and industrial sectors; agricultural and other treatments are often protective as well as curative; the recent onset of soybean rust has resulted in many new fungicides being registered and used; fungicides are used on an increasingly wide variety of vegetable, fruit, and grain crops; recent usage has been related to increased yields in otherwise healthy crops; and fungicide resistance will continue to challenge agriculture and impact usage rates and patterns sometimes resulting in the need for larger doses to kill resistant organisms and other times restricting or eliminating the use of a fungicide.

1.3 Environmental Fate

Characteristics of the different fungicides determine how they are used in a pest management program (Tenuta et al. 2007). Protective fungicides, such as chlorothalonil, pyraclostrobin, and azoxystrobin, prevent fungi from successfully colonizing on host tissue, but have little effect on colonies already established on host tissue. Thus, these fungicides would be applied before the onset of soybean rust for preventative purposes. Curative fungicides, such as propiconazole, tebuconazole, and myclobutanil, can inhibit or stop existing fungal infections by limiting spore germination, helping slow disease development. Thus, these fungicides would be applied after fungi have successfully colonized on a plant to cure or limit further development of such fungal infections.

An increase in fungicide use either in response to diseases such as soybean rust, to increased yield potential, or any of the other reasons noted above may lead to increased transport of fungicides and fungicide TPs to US water resources. Fungicides have a wide range of physical properties which affect their occurrence and fate in the environment (supplemental material, Table S1). For example, azoxystrobin or propiconazole are expected to be mobile in the environment because they are water soluble, slow to degrade in soil and water, unlikely to volatilize, and unlikely to partition to soils or sediment in aquatic systems. Other fungicides such as chlorothalonil or pyraclostrobin, are not as likely to be mobile in the environment because they are less water soluble, degrade more rapidly in soil and water, and would be expected to partition rapidly to soils or sediment in aquatic systems. In addition, fungicides can enter streams by way of wastewater discharge or storm water runoff in urbanized areas (Kahle et al. 2008).

Few studies have investigated the occurrence of fungicides in surface water. Studies in agricultural regions of Germany, however, found fungicides to be commonly present in streams with concentrations up to 30 μg/L (Berenzen et al. 2005; Liess and Von Der Ohe 2005). In samples from Swiss Midland lakes, fluconazole, propiconazole, and tebuconazole were frequently detected at with concentrations up 0.009 μg/L (Kahle et al. 2008). In samples from select US streams chlorothalonil was infrequently detected, whereas its major transformation product 4-hydroxy-chlorothalonil was detected more frequently (Scribner et al. 2006; Battaglin et al. 2008).

The potential water quality effect of fungicide use to combat soybean rust in Indiana was recently assessed using a modeling approach (Deb et al. 2010). Under a worst-case scenario, they simulated edge-of-field concentrations of 14 fungicides and compared them with concentrations of concern for chronic exposure to humans from drinking water and to fish in aquatic habitats (Plotkin et al. 2009a, 2009b). They concluded that most fungicides had the potential to occur in edge-of-field runoff at greater than 1 μg/L concentrations, and that several, including chlorothalonil, pyraclostrobin, and tebuconazole, had the potential to occur in runoff at concentrations that are above their chronic exposure thresholds (Deb et al. 2010).

1.4 Effects of Fungicides in the Environment

Fungicides are used to kill fungi and (or) their spores. Fungicides can be chemical or biological and function systemically or via contact with plant surfaces. All fungicides are at risk of losing all or part of their effectiveness due to development of fungicide resistant fungi (Brent and Hollomon 1998). Modes of action, whether the product functions systemically or on contact, the potential for onset of resistance, compound-specific behavior in the environment, and other factors all are taken into account by manufacturers when providing guidance on use of fungicides and related formulations. In turn, this accounts for the amount and timing of fungicide use and related pathways to the environment.

In the environment, fungicides can be toxic to fish, invertebrates, and other non-target organisms at a wide range of concentrations. Most toxicity estimates from the US Environmental Protection Agency’s ECOTOX database (US Environmental Protection Agency 2009a), and many from other published sources (Bartlett et al. 2002; Belden et al. 2007; Berenzen et al. 2005; Birge et al. 2000; Liess and Von Der Ohe 2005; Ochoa-Acuna et al. 2009; Plotkin et al. 2009a, 2009b) are 50th-percentile effect concentration or 50th-percentile lethal concentrations. Therefore, confounding factors such as effects of mixtures, novel or other modes of action (e.g., hormonally active processes), chronic exposures to non-target organisms as well as beneficial and indigenous fungal communities, and the potential for fungicide resistance patterns are not accounted for in present paradigms.

Toxicity estimates for the 12 fungicides (and some individual fungicides) assessed span more than four orders of magnitude from less than 10 to more than 10,000 μg/L (Fig. 2). In some cases, estimates of a fungicide’s toxicity to a class of organisms varied by more than two orders of magnitude (i.e., azoxystrobin and diatoms). The toxicity of some fungicides to Daphnia magna increased by an order of magnitude or more with an increase in exposure duration from 24 to 96 h (Ochoa-Acuna et al. 2009). The wide range of toxicity estimates for fungicides as well as the limitations they impose on interpretations of environmental occurrence data was not unexpected as similar results for other pesticides have been documented (Kerby et al. 2009).
https://static-content.springer.com/image/art%3A10.1007%2Fs11270-010-0643-2/MediaObjects/11270_2010_643_Fig2_HTML.gif
Fig. 2

Concentrations of 12 fungicides in 103 water samples collected in 2005 and 2006, and fungicide toxicity to selected aquatic organisms and humans

Toxicological data for amphibians and many other non-target aquatic species are limited. For example, no data were found that described the toxicity of the target fungicides analyzed for in this study to naturally occurring beneficial fungi such as those from the class Trichomycetes (Lichtwardt et al. 2001; White et al. 2006). Some fungicides exhibit endocrine-disrupting activities in mammals and several other toxic modes of action have been described (Zarn et al. 2003; Taxvig et al. 2007a, b; Maltby et al. 2009; Sancho et al. 2009). A recently conducted fungicide risk assessment (Maltby et al. 2009) indicated there was “...no appropriate toxicity data for fungi...”, but that there also was no evidence that toxicity estimates based on acute data from non-fungal species were not adequately protective. The target fungicides are similarly toxic to aquatic organisms as are the herbicides and insecticides that have traditionally been used in corn and soybean production (Battaglin and Fairchild 2002). Finally, some pesticide mixtures can have adverse effects on aquatic communities even when no individual compound occurs at toxic levels (Castillo et al. 2006; Hayes et al. 2006; Relyea 2009). The toxicological effect of mixtures of fungicides, like all pesticides, is poorly understood.

1.5 Objectives

The objective of this study was to collect baseline data on the occurrence of 12 target fungicides in select US streams soon after the first documentation of soybean rust in the US and prior to the corresponding increase in fungicide use to control and treat this problem. These data were used to identify spatial and seasonal patterns in fungicide occurrence and to investigate potential relations between fungicide occurrence and the use of those products in the associated drainage basins. This is the first study to look for azoxystrobin, boscalid, cyproconazole, metalaxyl, metconazole, pyraclostrobin, tetraconazole, and trifloxystrobin in streams across the US.

2 Methods and Materials

2.1 Study Sites and Sample Collection

Study sites in the southern and central US (Fig. 1) were selected on the basis of estimates of soybean acreage, expected fungicide use, watershed area, and the availability of ancillary data (such as streamflow). Our objective was to select sites with a range of drainage areas representing a range in hydrologic conditions, some of which have a substantial portion (more than 25%) of that drainage area in soybean production. Sites were also selected where fungicide use was (e.g., Georgia and Mississippi) and was not (e.g., Iowa) a common crop management practice.

A total of 103 water samples (not including quality assurance (QA) samples) from 29 streams across 13 states were collected between 2005 and 2006 (Table 1). A majority of sites (20 of 29) had multiple samples collected, with as many as four sample collected in a given year. Study sites ranged in drainage area from 61 to 221,000 km2. Drainage basin area in soybean production ranged from 0% to 41% (Table 1). Most samples were collected in summer and early fall during the time when fungicides would typically be applied for the prevention of soybean rust or be used on other crops, but some samples were collected as early as April or as late as November. A few samples were collected during runoff events, but these events were not specifically targeted for this study. All samples were collected using standardized protocols (Wilde et al. 1999) by wading or from bridges using the equal-width-increment method (Shelton 1994).
Table 1

Study sites for soybean rust study, 2005–2006 (km2)

Site no.

Station name

USGS station identification number

Drainage area (km2)

Percentage of drainage area in soybeans in 2002

Number of samples in 2005

Number of samples in 2006

Iowa

1

Turkey River at Garber, IA

05412500

4,002

21.7

1

1

2

Wapsipinicon River near De Witt, IA

05422000

6,050

28.9

1

1

3

Skunk River at Augusta, IA

05474000

11,170

28.6

1

1

4

Little Sioux River near Turin, IA

06607500

9,132

35.9

1

0

5

Boyer River at Logan, IA

06609500

2,260

36.7

1

0

6

S. Fork Iowa R. NE of New Providence, IA

05451210

580

36.1

1

3

7

Iowa River at Wapello, IA

05465500

32,380

32.4

0

2

Illinois

8

Sangamon River at Monticello, IL

05572000

1,420

41.0

1

4

Indiana

9

Sugar Creek at New Palestine, IN

394340085524601

240

36.5

0

3

Ohio

10

Auglaize River near Fort Jennings, OH

04186500

860

36.7

0

3

North Carolina

11

Neuse River at Kinston, NC

02089500

6,972

5.9

2

3

12

Contentnea Creek at Hookerton, NC

02091500

1,900

11.5

2

3

Nebraska

13

Maple Creek near Nickerson, NE

06800000

953

27.5

1

1

14

Platte River at Louisville, NE

06805500

221,000

3.9

0

3

Minnesota

15

Little Cobb River near Beauford, MN

05320270

337

31.8

1

4

Mississippi

16

Bogue Phalia nr Leland, MS

07288650

1,250

30.9

3

4

17

Yazoo R bl Steele Bayou nr Long Lake, MS

07288955

34,590

12.0

4

4

Georgia

18

Lime Creek near Cobb, GA

02350080

159

0.6

2

4

19

Sope Creek near Marietta, GA

02335870

76

0.0

2

1

20

Chattahoochee River near Whitesburg, GA

02338000

6,290

0.0

2

0

Alabama

21

Tombigbee R bl Coffeeville L&D near Coffeeville, AL

02469762

47,700

1.5

2

3

22

Flint River at Brownsboro, AL

03575100

971

3.9

1

3

23

Cahaba Valley Creek at Cross CR RD at Pelham, AL

0242354750

66

0.0

2

0

Kentucky

24

Bayou de Chien near Clinton, KY

07024000

178

34.7

1

1

Missouri

25

Locust Creek near Unionville, MO

06900900

201

10.2

1

0

26

Medicine Creek near Harris, MO

06899950

497

10.0

1

0

South Carolina

27

Gills Creek at Columbia, SC

02169570

154

0.8

3

1

28

Cow Castle Creek near Bowman, SC

02174250

61

2.9

2

4

29

Edisto River nr Givhans, SC

02175000

7,070

1.5

3

4

2.2 Analytical Methods

Several modifications (Sandstrom, oral communication) to an analytical method (Zaugg et al. 1995) were made to measure the concentrations of 11 fungicides at sub μg/L concentrations in filtered water samples. In 2005, the method was first modified to measure six fungicides (azoxystrobin, chlorothalonil, myclobutanil, propiconazole, pyraclostrobin, and tebuconazole) in water. In 2006, the method was further modified to include five more fungicides (boscalid, cyproconazole, metconazole, tetraconazole, and trifloxystrobin). Cis- and trans-propiconazole were analyzed separately but results were reported as the sum of these two isomers. Concentrations of one other fungicide (metalaxyl) were measured using another method (Sandstrom et al. 2001). The fungicides were isolated from 1-L water samples into a 0.5-g octadecyl-bonded silica solid-phase extraction column and eluted with 2 ml of ethyl acetate. After solvent evaporation and exchange to toluene, the extract was analyzed by gas chromatography, with selected ion-monitoring mass spectrometry. The laboratory reporting level (LRL; Childress et al. 1999) for the 11 fungicides ranged from 0.004 to 0.010 μg/L (Fig. 2).

Most (94 of 103) samples also were analyzed for 63 moderate use pesticides and pesticide TPs using solid-phase extraction and gas chromatography/mass spectrometry (Sandstrom et al. 2001), and selected samples (25) were analyzed for two chlorothalonil degradation products: 4-hydroxy-chlorothalonil and 1-amide-4-hydroxy-chlorothalonil using a method described in Scribner et al. (2006). From these analyses, only concentrations of the fungicide metalaxyl are included in this interpretation.

QA samples consisted of laboratory reagent blanks, laboratory reagent spikes, laboratory duplicates, field matrix spikes, field blanks, and field duplicates. There were no detections of target fungicides in the 27 laboratory reagent blanks or the ten field blanks. Eight field duplicates were collected allowing for a total of 91 possible comparisons of fungicides concentrations. The presence or absence of fungicides was confirmed in 98% of these comparisons. In two instances at concentrations near the reporting level, there was a detection of a fungicide in the field duplicate sample, but not in the corresponding environmental sample. The absolute difference in measured detections ranged from 0.004 to 0.108 μg/L, with the average difference being 0.024 μg/L.

Field and laboratory matrix spikes were prepared to evaluate the effects of sample matrix on fungicide recovery. In 2005, 16 matrix spike samples from 13 streams were analyzed. At three sites, duplicate matrix spikes were prepared, one spike being added at the collection site and one spike added at the laboratory. Calculated recoveries of fungicides in stream matrix spike samples are shown in Table S2. Calculated spike recoveries were not used in the 2005 summary statistics for three fungicides in one sample because the concentrations in the environmental samples were close to the spike concentration level (ratio of spike/environmental <1.5). In 2005, median recoveries of azoxystrobin (215%) and myclobutanil (177%) were found to be elevated (Table S2). This was in contrast to the recoveries of chlorothalonil (86%), cis-propiconazole (109%), and trans-propiconazole (90%) that were closer to 100%. The recoveries in spike samples indicate that some fungicides have a high bias from the sample matrix which could translate to a positive bias in the corresponding environmental samples. This, however, only effects measured concentrations and does not introduce false-positives in the data. Because the bias was higher in stream samples than in reagent water, the cause is most likely the stream sample matrix. Similar enhancement in observed chromatographic response was observed for pesticide residues in a matrix extract when compared with the same concentration in a matrix-free solution (Poole 2007). The matrix increases the transfer of fungicides from hot vaporizing injectors by reducing the thermal stress for labile compounds and by masking active sites in the injector responsible for the adsorption or decomposition of polar pesticides.

In 2006, 21 matrix spike samples from 12 streams were analyzed, 15 using a spike mixture that contained ten fungicides, and six using a spike mixture with four fungicides. All samples were spiked in the field into filtered water samples. Expected spike concentrations were 0.1 μg/L. Calculated recoveries of fungicides in stream matrix spike samples are shown in Table S2. Calculated spike recoveries were not used in the 2006 summary statistics for three samples because the concentrations in the environmental samples were close to the spike concentration level (ratio of spike/environmental <1.5). The calculated spike recoveries had high positive bias and were quite variable with median values for azoxystrobin (268%), boscalid (310%), and pyraclostrobin (399%; Table S2). These fungicides have chemical structures that cause them to be susceptible to thermal stress of the hot injector and active sites in the injector or gas chromatograph (GC) column. Co-extracted matrix components block active sites in the GC, reducing losses of these susceptible analytes, and results in higher analyte signals compared to the matrix-free calibration solutions (Mastovska et al. 2005).

The matrix spike recovery results indicate that true concentrations of azoxystrobin, boscalid, and pyraclostrobin in the streams might be two to five times lower than reported concentrations. However, the fungicide concentrations reported here were not corrected for enhanced recoveries because matrix spikes results were not available for every site and sample, and the sample matrix and resulting bias was expected to be different for each sample. The elevated spike recoveries have no impact on the detection of the target fungicides. Consequently, interpretation of concentration data is qualified and limited in this paper to general perspective on the range and maximum concentrations encountered. With detection frequencies these data provide adequate indications of the presence and magnitude of the target fungicides in the environments sampled.

In some cases, fungicide concentrations that were less that the LRL were reported. These values were flagged by the laboratory with an “E” code which indicates that the sample met all retention time and mass spectral acceptance criterion (Focazio et al. 2008). These values are displayed graphically and used for the calculation of detection frequency, but may not be appropriate for other statistical uses.

2.3 Fungicide Use Estimates

The total annual use of selected fungicides on all crops by county circa 2002 was calculated from data provided in Gianessi and Reigner (2006). An area-weighted sum algorithm programmed in a geographic information system (Battaglin and Goolsby 1998) was used to estimate the amounts of fungicides used or areas of soybeans harvested within drainage basins associated with the sampling sites. The algorithm accounts for cases where an entire county is within a single drainage basin and where only a portion of a county is within a drainage basin. The total annual use of 12 fungicides (Table S1) on all crops by state for 2002–2007 was prepared by the USGS using AgroTrak® agrochemical usage data (AgroTrak® is a registered trademark of dmrkynetec, St. Louis, MO, USA).

3 Results and Discussion

3.1 Fungicide Occurrence

Nine of the 12 fungicides analyzed were detected in at least one sample (Fig. 2, Table 2), at least one fungicide was detected in 58 (56%) of the 103 samples, and at least one fungicide was detected in 20 of 29 streams. Two or fewer samples were collected at eight of the nine streams with no fungicide detections. As many as five fungicides were detected in an individual sample, and mixtures of fungicides were common. Azoxystrobin was the most frequently detected fungicide (45%, 1.13 μg/L maximum concentration; Table 2) followed by metalaxyl (27%, 0.067 μg/L maximum concentration); propiconazole (reported as the sum of the concentrations of trans- and cis-propiconazole; 17%, 1.15 μg/L maximum concentration) and myclobutanil (9%, 0.032 μg/L maximum concentration). Tebuconazole, pyraclostrobin, trifloxystrobin, and chlorothalonil were infrequently detected in this study. Boscalid, cyproconazole, and metconazole were not detected in any samples.
Table 2

Summary of water quality analysis for selected fungicides, chlorothalonil degradation products in selected US streams 2005–2006 (μg/L)

Fungicide

Number of samples

Number of detections

Number of sites with detections (out of 29 possible)

Laboratory reporting level in μg/L

Mean of detections in μg/L

Maximum concentration in μg/L

Azoxystrobin

103

46

17

0.008

0.163

1.13

Boscalid

103

0

0

0.006

<0.006

<0.006

Chlorothalonil

103

2

2

0.004

0.031

0.033

Cyproconazole

103

0

0

0.006

<0.006

<0.006

Metalaxyl

94

25

10

0.007

0.017

0.067

Metconazole

103

0

0

0.006

<0.006

<0.006

Myclobutanil

103

9

8

0.004

0.012

0.032

Propiconazole

103

18

4

0.004

0.291

1.15

Pyraclostrobin

103

3

2

0.006

0.031

0.054

Tebuconazole

103

6

3

0.010

0.053

0.115

Tetraconazole

103

1

1

0.006

0.047

0.047

Trifloxystrobin

103

2

2

0.004

0.029

0.029

4-hydroxy-chlorothalonil

25

0

0

0.050

<0.050

<0.050

1-amide-4-hydroxy-chlorothalonil

25

0

0

0.050

<0.050

<0.050

Azoxystrobin was found in samples from 17 sites in 11 states (Fig. 3a) with 11 of these sites having detections in more than one sample. Its relatively widespread occurrence may be due to its high water solubility and its rapidly increasing use in a wide variety of agricultural and non-agricultural settings (US Environmental Protection Agency 1997; Gianessi and Reigner 2006; US Department of Agriculture 2009b). The maximum concentration of azoxystrobin (1.13 μg/L) from this study (Table 2) was less than that reported in German streams (11.1 and 29.7 μg/L; Berenzen et al. 2005). Azoxystrobin was detected in one sample from Indiana, where simulation models suggested that it could occur in runoff there under a worst-case scenario (Deb et al. 2010). There was significant direct correlation (Spearman’s rho) between the maximum azoxystrobin concentration at a site (Fig. 3a) and estimates of the use of azoxystrobin in the upstream drainage basins (p = 0.007).
https://static-content.springer.com/image/art%3A10.1007%2Fs11270-010-0643-2/MediaObjects/11270_2010_643_Fig3_HTML.gif
Fig. 3

Estimated 2002 annual use rate and the potential daily flux for: a azoxystrobin, b myclobutanil, c propiconazole, and d tebuconazole

Metalaxyl was found in samples from ten sites in eight states (Alabama, Indiana, Iowa, Minnesota, Mississippi, North Carolina, Ohio, and South Carolina). The maximum concentration of metalaxyl from this study was 1.15 μg/L (Table 2). Metalaxyl use in US agriculture was not quantified for 2002 (Gianessi and Reigner 2006), but use in 1997 was estimated to be ∼300 metric tons. Only 0.8 metric ton of metalaxyl use in the US was reported by AgroTrak® in 2006. Metalaxyl is used alone or in mixtures with other fungicides as seed treatments for a wide variety of crops including soybeans, corn, and vegetables. Metalaxyl has many non-agricultural uses such as on turf and non-food crops, for which use estimates are not available.

Myclobutanil was found in samples from eight sites in five states (Fig. 3b). Myclobutanil was detected at least once at both sites from Mississippi and at all three sites in South Carolina. Myclobutanil was detected more than once in a year at only one site, the Neuse River, NC, USA. There was no correlation between the maximum myclobutanil concentration at a site (Fig. 3) and estimates of the use of myclobutanil in the upstream drainage basins (p = 0.58).

Cis- or trans-propiconazole was found in samples from four sites in three states (Fig. 3c). The maximum concentration of propiconazole (1.15 μg/L) from this study (Table 2) was greater than that reported for Swiss lakes (0.0019 μg/L; Kahle et al. 2008). There was significant direct correlation between the maximum propiconazole concentration at a site (Fig. 3) and estimates of the use of propiconazole in the upstream drainage basins (p = 0.047).

Tebuconazole was found in samples from three sites in two states (Fig. 3d). The maximum concentration of tebuconazole (0.115 μg/L) from this study (Table 2) was less than that reported in German streams (9.1 μg/L; Berenzen et al. 2005), but greater than that reported for Swiss lakes (0.001 μg/L; Kahle et al. 2008). Tebuconazole was detected more than once in a year at only one site, Bogue Phalia, MS, USA. There was no correlation between the maximum tebuconazole concentration at a site (Fig. 3) and estimates of the use of tebuconazole in the upstream drainage basins (p = 0.66).

Chlorothalonil was only detected in two samples from two sites in Georgia even though it was used in all 13 states, and the rate of its use was much greater than for the other fungicides (Table S1; Gianessi and Reigner 2006). Chlorothalonil physical properties and degradation characteristics are such that it is seldom observed in surface-water samples (Battaglin et al. 2008; Scribner et al. 2006). At one of the sites, Lime Creek, GA, USA the estimated use rate (total use in the basin divided by drainage area) of chlorothalonil was much larger (20,220 gm/km2) than at any other site. The use rate was only 5.2 gm/km2 at the other site with a detection of chlorothalonil (Sope Creek, GA, USA). The two degradation products of chlorothalonil, 4-hydroxy-chlorothalonil and 1-amide-4-hydroxy-chlorothalonil, were not detected (LRL of 0.050 μg/L) in any of the 25 samples that were analyzed.

Boscalid, cyproconazole, and metconazole were not detected in any sample. Boscalid was first registered for use in the US in 2003 and is used on a wide variety of vegetable crops but not significantly on soybeans (US Department of Agriculture 2009b). Some boscalid was used in Georgia, Illinois, Indiana, Minnesota, Missouri, and North Carolina in either 2005 or 2006. Cyproconazole was first registered for use in the US in 1988 as a wood preservative. It is used in some turf products and was registered for use on soybeans in 2008. There was no reported cyproconazole use in 2005 or 2006 in any of the 13 states where samples were collected (Fig. 1). Metconazole was not registered for use on crops in the US in 2005 and 2006 and is used internationally primarily on bananas (US Environmental Protection Agency 2006a). Metconazole is currently (2009) approved for treatment of soybean rust but is not being marketed, and there was no reported use in 2005 or 2006 (AgroTrak®) in any of the 13 states where samples were collected.

3.2 Relations to Toxicity and Use

The concentrations of fungicides in US streams measured in this study were in general one or more orders of magnitude less than toxicity estimates for those individual compounds on aquatic organisms or humans (Fig. 2). Fungicides, however, appear to frequently contribute to the mixture of pesticides that aquatic organisms in these habitats can be exposed to. Because fungicides typically are applied later in the growing season than are herbicides, their use in areas that historically have not had fungicide applications means that non-target organisms are now potentially exposed via new pathways to these compound. Other confounding factors such as effects of mixtures, novel or other modes of action (e.g., hormonally active processes), chronic exposures to non-target organisms as well as beneficial and indigenous fungal communities, and the potential for fungicide resistance patterns are not accounted for in present toxicological benchmarks. Therefore, the comparison of measured fungicide concentrations with current toxicological benchmarks provides some perspective on potential environmental hazards many potential impacts remain incompletely quantified.

The lack of correlation between the use and occurrence of some fungicides is likely the result of two factors: (1) insufficient number of samples to accurately characterize fungicide occurrence and (2) a lack of accurate fungicide use data by county on a national scale for 2005 or 2006. In particular, information on the use of newer fungicides is out-of-date at best (circa 2002), and does not account for emergency exemptions or non-agricultural use.

State-level fungicide use estimates from AgroTrak® confirm that the circa 2002 county-level data (Gianessi and Reigner 2006) may not be sufficient for identification of relations between fungicide occurrence and use in 2005 and 2006. For example, azoxystrobin use is reported by AgroTrak® in both 2005 and 2006 for all 11 states with reported detections including Iowa and Nebraska, which do not show use in the circa 2002 data (Fig. 3a). Likewise, myclobutanil and tebuconazole were detected at both sites in Mississippi and use is reported by AgroTrak® in 2005 in Mississippi when none is shown in the 2002 county-level data (Fig. 3b, d). Pyraclostrobin was detected in three samples from two sites: Bogue Phalia, MS, USA and Bayou De Chien, KY, USA. Gianessi and Reigner (2006) do not indicate any use of pyraclostrobin in 2002 in Mississippi or Kentucky; however, the AgroTrak® data indicate that there was use of pyraclostrobin in both states starting in 2003. Tetraconazole was detected in one sample from the Iowa River, IA, USA. Gianessi and Reigner (2006) do not indicate any use of tetraconazole in 2002 in Iowa; however, the AgroTrak® data indicate that there was use of tetraconazole in Iowa in 2005.

3.3 Seasonal Occurrence

At seven sites, four samples were collected during 2005 or 2006. Results from these sites (Table 1; one site had four samples in both 2005 and 2006) were used to determine if there is a seasonal pattern to fungicide occurrence or concentration as previously documented for herbicides (Thurman et al. 1991; Battaglin et al. 2005; Scribner et al. 2005). At the three sites in warmer climates with moderate to high soybean production (Yazoo River, Bogue Phalia, and Cow Castle Creek) fungicides are detected throughout the growing season (Fig. 4a) with the highest concentrations occurring in July, August, or September (Fig. 4b). At the more northern sites (Sangamon River and Little Cobb River) and at sites with little soybean production (Lime Creek and Edisto River) fungicide detections are more sporadic but generally more common and at greater concentration during July, August, and September (Fig. 4).
https://static-content.springer.com/image/art%3A10.1007%2Fs11270-010-0643-2/MediaObjects/11270_2010_643_Fig4_HTML.gif
Fig. 4

The a number of target fungicides detected and b total concentration of target fungicides, by month for selected site in 2005 and 2006

Results from these sites suggest that in many US streams one should expect a seasonal pattern to fungicide occurrence, and that detection of fungicides is more likely in late summer and early fall than in spring. However, fungicides may occur throughout the growing season in agriculturally impacted streams in warmer climates suggesting the potential for season-long exposure of aquatic organisms to fungicide in these watersheds.

4 Summary

Results from this study indicate that fungicides occurred in streams with moderate frequency prior to their widespread use to control soybean rust or increase crop yields in the southern and central US. This study collected baseline data on the occurrence of 12 fungicides in 29 US streams soon after the first documentation of soybean rust in the US and prior to the corresponding increase in fungicide use to control and treat this problem. These data were used to assess spatial and seasonal patterns in fungicide occurrence and to investigate relations between the occurrence of fungicides in US streams and the use of those products in the associated drainage basins. Nine of the 12 fungicides were detected in at least one sample. At least one fungicide was detected in 56% of the samples and in 20 of 29 streams. As many as five fungicides were detected in an individual sample, and mixtures of fungicides were common. Azoxystrobin was detected most frequently (45% of samples) followed by metalaxyl (27%), and propiconazole (17%). Myclobutanil, tebuconazole, pyraclostrobin, trifloxystrobin, chlorothalonil, and tetraconazole were infrequently detected during this study. Boscalid, cyproconazole, and metconazole were not detected in any samples. Fungicide concentrations ranged from 0.002 to 1.15 μg/L. Generally, fungicides occurred more frequently and in higher concentrations in the southern US where their use on soybeans and other crops was more common and widespread, but detections also occurred where fungicide use on soybeans was not reported and fungicide use on other crops was not common suggesting that non-agricultural uses are a potential source of fungicide to the environment. The concentrations of fungicides in US streams measured in this study were in general one or more orders of magnitude less than toxicity estimates for those individual compounds on aquatic organisms or humans; however, there are limitations in available toxicological benchmarks. There is some evidence of a seasonal pattern to fungicide occurrence, with detections more likely and concentrations higher in late summer and early fall than in spring. At a few sites in Mississippi and South Carolina, fungicides were detected throughout the growing season. Fungicide occurrence in US streams appear to be positively correlated to the use of those products in the associated drainage basins; however, more detailed monitoring data and more accurate use information are needed to quantify these relations.

It is not possible to determine if any of the fungicide detections from this study were explicitly the result of treatments to cure or prevent soybean rust, because all of the detected compounds have more than one crop on which they are used and more than one disease that they treat. It is clear that some of the detections were in areas where soybean rust eventually spread to in either 2005 or 2006. However, these results show that many of the target fungicides are being introduced to the aquatic environment in low-level mixtures. It is also clear that fungicide occurrence will likely increase in the environment in response to increases in fungicide applications to prevent fungal disease outbreaks such as soybean rust and to increase crop yields. Finally, because this occurrence assessment was done prior to any major outbreak of soybean rust in the Central US, the results will provide powerful comparative information for future studies.

Acknowledgments

This study was supported by the USGS Toxic Substances Hydrology Program and conducted with the logistical support of USGS Water Science Centers in Alabama, Georgia, Illinois, Indiana, Iowa, Kentucky, Minnesota, Mississippi, Missouri, Nebraska, North Carolina, Ohio, and South Carolina. The authors would especially like to acknowledge USGS personnel who collected samples: Richard Coupe, Angela Crain, David Dupre, James Fallon, Jeffery Frey, Thomas Harris, Brian Hughes, Steve Kalkhoff, Michael Woodside, and Ron Zelt. The use of trade names or product names in this report does not constitute endorsement or recommendation by the USGS.

Supplementary material

11270_2010_643_MOESM1_ESM.doc (38 kb)
Table S1First used in US, crops applied to, physical properties, and amount used in agriculture in the US in 2006 for selected fungicides (Bartlett et al. 2002; US Environmental Protection Agency 1994, 1997, 1999, 2003, 2005, 2006a, b; DOC 38 kb)
11270_2010_643_MOESM2_ESM.doc (42 kb)
Table S2Recovery of fungicides, in percent, in field matrix spike samples fortified at a concentration of 0.1 μg per liter, analyzed in 2005 and 2006 [−, no samples] (DOC 41 kb)

Copyright information

© Springer Science+Business Media B.V. (outside the USA) 2010