COMPARISON OF PESTICIDES IN EIGHT U.S. URBAN STREAMS

Environmental Toxicology and Chemistry, Vol. 19, No. 9, pp. 2249–2258, 2000 Printed in the USA 0730-7268/00 $9.00 1 .00

COMPARISON OF PESTICIDES IN EIGHT U.S. URBAN STREAMS RYAN S. HOFFMAN,† PAUL D. CAPEL,*‡ and STEVEN J. LARSON‡ †Department of Geology and Geophysics, University of Minnesota, Minneapolis, Minnesota 55455, USA ‡U.S. Geological Survey, Department of Civil Engineering, University of Minnesota, Minneapolis, Minnesota 55455 ( Received 10 June 1999; Accepted 21 January 2000) Abstract—Little is known of the occurrence of pesticides in urban streams compared to streams draining agricultural areas. Water samples from eight urban streams from across the United States were analyzed for 75 pesticides and seven transformation products. For six of the eight urban streams, paired agricultural streams were used for comparisons. The herbicides detected most frequently in the urban streams were prometon, simazine, atrazine, tebuthiuron, and metolachlor, and the insecticides detected most frequently were diazinon, carbaryl, chlorpyrifos, and malathion. In contrast to similar-sized agricultural streams, total insecticide concentrations commonly exceeded total herbicide concentrations in these urban streams. In general, the temporal concentration patterns in the urban streams were consistent with the characteristics of the local growing season. The insecticides carbaryl and diazinon exceeded criteria for the protection of aquatic life in many of the urban streams in the spring and summer. When the country as a whole is considered, the estimated mass of herbicides contributed by urban areas to streams is dwarfed by the estimated contribution from agricultural areas, but for insecticides, contributions from urban and agricultural areas may be similar. The results of this study suggest that urban areas should not be overlooked when assessing sources and monitoring the occurrence of pesticides in surface waters. Keywords—Pesticide

Herbicide

Insecticide

Urban

Stream

pesticides in surface waters have been well characterized for agricultural settings [8–10]. Only limited work, however, has been done on the occurrence of pesticides in urban surface waters [11–14]. The U.S. National Urban Runoff Program study [12] represents the only previous attempt to quantify the extent and frequency of occurrence of pesticides in urban runoff on a national scale. Samples in the U.S. Urban Runoff Program study were collected during isolated storm-water runoff events, and temporal patterns were not considered. Most of the pesticides analyzed were organochlorine insecticides, and most of these are now banned in the United States. The results indicated that these pesticides were not present extensively in the nation’s streams at the quantification levels of the U.S. Urban Runoff Program study. Oltmann and Shulters [13] measured pesticides in rain and runoff from various urban settings (commercial, residential, and industrial) in Fresno, California. They detected the insecticides diazinon and malathion in almost all rain and runoff samples. Many of the other pesticides for which their samples were analyzed were organochlorine insecticides no longer used in the United States. Wotzka et al. [14] monitored the occurrence of a variety of urban and agricultural pesticides entering an urban lake in Minneapolis, Minnesota. They found that 85% of storm runoff samples contained commonly used urban herbicides (2,4-D, dicamba, MCPP, and MCPA) and that 43% of storm runoff samples contained commonly used agricultural herbicides (alachlor, atrazine, cyanazine and metolachlor). On the basis of data from rainfall samples collected during the same period, the authors concluded that the agricultural herbicides were introduced into the watershed via atmospheric deposition. A study of pesticides in the Mississippi River Basin has shown that urban use of diazinon results in measurable concentrations in several major rivers [9]. For the most part, however, the significance of urban areas as a source of pesticides

INTRODUCTION

Pesticides are used to control weeds and insects in both agricultural and urban settings. Urban pesticide uses include, but are not limited to, lawns and gardens, structures, mosquito control, golf courses, roadsides, and pet shampoos. Nonagricultural uses account for approx. 25% of total pesticide use in the United States, with a substantial portion of this use in urban areas [1]. Aspelin [1] reported that home and garden pesticide use has remained relatively constant between 1979 and 1995 at about 30 to 40 million kg of active ingredient per year. These values include both indoor and outdoor uses of herbicides, insecticides, and fungicides. Barbash and Resek [2] reported that insecticide use on lawns and golf courses is about 240 and 1,300 kg active ingredient/km2, respectively. They also report that herbicide use on lawns and golf courses is about 580 and 500 kg active ingredient/km2, respectively. These application rates are considerably higher than the rates reported for agricultural applications. The average application rates on cropland in the conterminous United States that is treated with insecticides and herbicides are estimated as 140 and 90 kg active ingredient/km2, respectively [3]. Sales of lawn and garden pesticides, which are applied by both home owners and professional applicators, were estimated at $1.1 billion in 1991 [4]. Few surveys have been conducted to evaluate pesticide use in urban areas. These include surveys of homeowners [5,6] and professional pesticide applicators [7]. In each of these surveys, it was found that insecticides were used more frequently than herbicides. The occurrence, concentrations, and seasonal patterns of * To whom correspondence may be addressed ([email protected]). The use of firm, trade, and brand names is for identification purposes only and does not constitute endorsement by the U.S. Geological Survey. 2249

Environ. Toxicol. Chem. 19, 2000

For Las Vegas population density and percentage urban land use are based on urban land-use area of the urbanized portion of the watershed, not total basin area, because a large theoretical portion of the watershed is in a scarely populated desert area.

OR

NY

NV

GA

FL

DC

CO

a

16 5.1 46 150 8.3 150 180 1,200 7 66 2 99 77 1 92 1 40 155 80 39 525 18 1,200 55 20,961 2,796 96,508 2,134

160 11 170 6.1 31 5.3 31 4.4 6.2 81 230 320 460 82 80 200 46 40 2 33 0 77 0 56 0 67 0 96 1 98 3 92 2 98 0 80 61 1,473 60 457 25 420 80 161 2,641 1,830 2 1,423 66 674 9 791 20 2,010 111,912 2,543 85,667 30,120 16,955 3,807 63,151 3,298 472,393

22 49 4 50 85 253

Total insecticide yield (g/year/km2) Total herbicide yield (g/year/km2) Agricultural land use (as % of basin) Urban land use (as % of basin) Total basin area (km2) 1990 basin population density (1/km2) 1990 basin population

21,498

Norwalk River, Winnipauk, CT — Cherry Creek, Denver, CO Lonetree Creek, Greeley, CO Accotink Creek, Annandale, VA Monocacy River, Bridgeport, MD Lafayette Creek, Tallahassee, FL Tucsawhatchee Creek, Hawkinsville, FL Sope Creek, Marietta, GA Lime Creek, Cobb, GA Las Vegas Wash, Las Vegas, NVa — Lisha Kill, Niskayuna, NY Canajoharie Creek, Canajoharie, NY Fanno Creek, Durham, OR Zollner Creek, Mt. Angel, OR

Samples were collected using the equal-width-increment sampling method into a Teflont sampler bottle [17]. After the sample was collected, a Teflon cone splitter was used to obtain two 1-L aliquots for pesticide analysis. Each aliquot was filtered through a 142-mm glass-fiber filter with nominal 0.7mm pore openings. Only the filtrate was analyzed for the pesticides. Based on their sorptive characteristics, most of the pesticides exist mainly in the aqueous phase. Only a few compounds (chlorpyifos, p,p9-DDE, pendimethalin, cis-permethrin, and propargite) are sorptive enough (log Koc $ 4) to po-

CT

Sample collection

Urban stream, location Agricultural stream, location

METHODS

General designation

to surface waters is difficult to determine. Many of the compounds used in urban areas are also used in agriculture (such as 2,4-D, dicamba, trifluralin, chlorpyrifos, and diazinon), so that the source of these pesticides detected in surface waters is often unclear. In addition, pesticides used almost exclusively in urban areas (such as isazofos, isophenphos, oryzalin, and MCPP) have seldom been targeted in studies of surface-water quality. Processes affecting the movement of pesticides to surface waters in urban areas are the same as those in agricultural areas, but some important differences between the two environments could affect this movement. Urban areas have large expanses of impermeable surfaces, such as concrete and asphalt roads, parking lots, and sidewalks, from which pesticides can be easily removed by runoff water from rain or sprinklers. These surfaces also provide a more or less continuous pathway along which pesticides may be transported by water. Thus, if pesticides applied in urban areas reach impervious surfaces (via spray drift, direct aerial application, misdirected application, or runoff from lawns and gardens), there is a relatively high probability that they will be transported to surface-water bodies. In studies done with turf plots, however, it has been found that very little runoff occurs from well-maintained grass, even with large amounts of precipitation [15]. Thus, at least for applications to well-maintained lawns, the limiting step may be reaching the impervious surfaces. Storm sewer systems also provide a direct pathway for movement of pesticides to lakes or rivers. Similarly, effluent from sewage treatment plants may contain pesticides, particularly in urban areas in which storm sewers and sanitary sewers are combined. The objective of this paper is to compare the occurrence of current-use pesticides in eight urban streams in diverse areas of the United States (Table 1). These data were collected during 1993 and 1994 as part of the U.S. Geological Survey’s National Water-Quality Assessment Program [16]. This is the first study of the occurrence of pesticides in urban environments across the nation since the U.S. Urban Runoff Program study [12] and the first to target currently used pesticides. In particular, this paper addresses the occurrence, spatial and temporal patterns, basin yields, and potential sources of 75 pesticides and seven pesticide transformation products (Table 2). The names of the sampled streams are given in terms of the large urban area in or near which the sampling site is located (Table 1). A few of the sampling sites are within large cities: Tallahassee, Florida; Denver, Colorado; and Las Vegas, Nevada. The remaining sites are in the suburban areas of large cities: Annandale, Virginia (near Washington, DC); Durham, Oregon (near Portland, OR); Marietta, Georgia (near Atlanta, GA); Niskayuna, New York (near Albany, NY); and Winnipauk, Connecticut (near Norwalk, CT).

R.S. Hoffman et al.

Table 1. Description of the urban and paired agricultural stream basins and their yields of herbicides and insecticides

2250

Environ. Toxicol. Chem. 19, 2000

Comparison of pesticides in eight urban streams

2251

Table 2. Summarized occurrence of pesticides and transformation products with their common uses

Common name Herbicides Simazine** Prometon** Atrazine** Tebuthiuron** Metolachlor** 2,4-D (acid)* Diuron* Pendimethalin** DCPA** Metribuzin** Oryzalin* MCPA* Triclopyr* Alachlor** Dichlobenil* Cyanazine** Dichlorprop* S-ethyl dipropylthiocarbamate (EPTC)** Napropamide** Terbacil** Benfluralin** 2,4-DB ester* Bromacil* Molinate** Propanil** Propham* Triallate** Trifluralin**

Common uses1,2

A, R A, A, A A, A, A, A, A A, A, A, A A, A A, A A A A, A A, A A A, A A,

Q, T T R T O O, T O, T O, T T R, T T Q, R, T

T R T O

Insecticides Diazinon** Carbaryl** Chlorpyrifos** Malathion** Dieldrin**,5 Fonofos** Carbofuran** Propoxur* Methomyl* g-HCH (Lindane)** Methyl Azinphos** Parathion** Propargite**

A, A, A, A, A A, A, A, A, A, A, A A

Transformation products1 Desethylatrazine**

Atrazine

S, T O, T P, S O, P O O S O P O

Detection frequency (%)4

Maximum concentration (mg/L)

71.6 70.7 54.0 21.9 19.5 12.5 10.0 9.8 6.5 3.7 3.5 3.0 2.5 2.3 1.4 1.4 1.0

8.2 2.93 0.755 0.14 0.49 1.2 6.2 0.32 0.045 0.09 1.9 1.3 0.34 0.052 0.3 0.045 0.19

All All All CO, GA, NV, OR CO, DC, FL, GA, NV, NY, OR CO, DC, GA, NY, OR DC, NV, OR CO, DC, GA CO, DC, NV, OR CO, DC, FL, NY, OR CO, DC, GA DC, GA DC, GA, OR CT, DC, GA, NV OR CO CT, DC

— — — — — 5C — — — 0.24C 0.2C

0.9 0.9 0.9 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5

0.037 0.021 0.035 0.011 0.39 0.06 0.016 0.022 0.07 0.036 0.011

CO, OR CT, GA CO, OR CT, DC GA GA CO, NY CO CO CO CT

0.08I 0.2C 0.041U 0.1U 0.56U — 1.8C — — 0.01I 0.01U 0.013U —

69.3 43.7 17.7 14.0 1.4 1.4 0.9 0.5 0.5 0.5 0.5 0.5 0.5

1.4 3.2 0.3 0.41 0.026 0.084 0.027 0.26 0.14 0.013 0.171 0.014 0.015

All CO, CO, CO, CO, DC DC, GA NV FL OR CO CO

3.3

0.04

CO, DC, OR

Water-quality criterion3

10C — 1.8C 1.6C 7.8C 4C — — — 1C — 2.6C — — — 2C —

—

Sites with detections (See Table 1 for site code)

CT, DC, FL, GA, NY, OR DC, FL, GA, NV, OR CT, DC, FL, GA, NV, OR DC, NV OR

Pesticides analyzed for but never detected: herbicides (common uses; water-quality criterion): 2,4,5-T*,5(A,R; —), Acifluorfen*(A,T; —), Bentazon*(A,T; —), Bromoxynil*(A,T; 5C), Butylate**(A; —), Chloramben*(A; —), Clopyralid*(A,O; —), DCPA(mono acid)*(A; —), Dicamba*(A,R,T; 10C), Dinoseb*,5(A; 0.05C), 4,6-Dinitro-o-cresol (DNOC)*,4(A; —), Ethalfluralin**(A; —), Fenuron*(A; —), Fluometuron*(A; —), Linuron**(A,R; 7C), MCPB*(A; —), Neburon*(A; —), Norflurazon*(A; —), Pebulate**(A; —), Picloram*(A,R; 29C), Pronamide**(A,O,T; —), Propachlor**(A; —), Silvex*,5(A,T; —), Thiobencarb**(A; —); insecticides: Aldicarb*(A; 1C), Dimethoate** (A,O; —), Disulfoton**(A,O; —), Ethoprop**(A,T; —), Methiocarb*(A,O,T; —), Methyl Parathion**(A; —), Oxamyl*(A,O; —), cis-Permethrin**(A,O,S,T; —), Phorate**(A; —), Terbufos**(A; —), transformation products1: Aldicarb Sulfone*(Aldicarb; 1 C), Aldicarb Sulfoxide*(Aldicarb; 1 C), 2,6-Diethylanaline**(Alachlor; —), a-HCH**(Lindane; —), 3-Hydroxycarbofuran*(Carbofuran; —), p,p9-DDE**(DDT; —)2 [3,32,33]. The transformation products are related to their parent compounds. In the case of a-HCH, it is both a transformation product and a component of the technical mixture of Lindane. 2 Abbreviations: A 5 agricultural; O 5 ornamental and garden; P 5 pets; Q 5 aquatic control; R 5 roadways, railways, and rights of way (includes brush control); S 5 structural (includes indoor and outdoor use); and T 5 turfgrass. 3 Water-quality criteria for the protection of aquatic health in mg/L: C 5 Canada [27,28], I 5 International Joint Commission [29], U 5 United States [26], and — 5 no criterion. 4 Reporting limits are either 0.05 mg/L or 0.01 mg/L, denoted by * or **, respectively. Detection frequency based on 200, 201, or 211 observations for compounds with reporting limits of 0.05 mg/L and 215 observations for compounds with reporting limits of 0.01 mg/L. For comparison, if all the data were censored at 0.05 mg/L, the detection frequencies of the following pesticides would have been: Simazine 37.6, Prometon 23.1, Atrazine 12.2, Tebuthiuron 2.7, Metolachlor 3.2, Diazinon 24.9, Carbaryl 21.3, Chlorpyrifos 0.9, and Malathion 5.45%. 5 Registration canceled or sales discontinued in the United States before the sampling period. 1

2252

Environ. Toxicol. Chem. 19, 2000

R.S. Hoffman et al.

tentially have a low bias in their concentrations due to the sampling procedure. Frequency of sampling varied from site to site and depended on the time of year. In general, sampling was more frequent during May through September (approximately four to eight samples per month), when pesticide use was generally the greatest. Sampling was less frequent during the rest of the year (approximately one to two samples per month), when pesticide use was generally lower in most parts of the country. Fewer samples were collected from the stream in Portland, Oregon (about 16 per year) than at the other sites. Daily streamflow data were obtained at each site.

Sample analysis The 75 pesticides and seven transformation products included in this study (Table 2) were determined by two methods. About half the pesticides were determined by capillary gas chromatography–mass spectrometry (GC/MS) [18], and the remaining pesticides were determined using high-performance liquid chromatography (HPLC) [19]. Pesticides determined using GC/MS were isolated on a C-18 solid-phase extraction column. The sorbed pesticides were eluted from the solidphase extraction column using a 3:1 hexane:isopropanol (v/v) solvent mixture. Selected-ion monitoring was used for detection. Recoveries for spiked reagent water samples ranged from 37 to 126% with a mean recovery of 83% [18]. Pesticides determined using HPLC were isolated by Carbopak-B solidphase extraction (Carbopak, Tekmann-Dohrman, Cincinnati, OH, USA). The basic and neutral compounds were eluted using a 4:1 methylene chloride:methanol (v/v) solvent mixture. The acidic compounds were eluted using a similar solvent mixture that was acidified with 0.2% trifluoroacetic acid. Photodiodearray detection was used in conjunction with HPLC. Recoveries for spiked reagent water samples ranged from 29 to 94% with a mean recovery of 63% [19]. The concentrations reported were not corrected for recovery. Because most of the pesticides analyzed by GC/MS had a method detection limit at or below 0.01 mg/L, the reporting limit was chosen as 0.01 mg/L for these compounds. Most of the pesticides analyzed by HPLC had a method detection limit at or below 0.05 mg/L, so the reporting limit of 0.05 mg/L was chosen for these compounds. These reporting limits were used in calculating detection frequencies (Table 2). Comparisons among the pesticides analyzed by GC/MS and HPLC are made in this paper; thus, the reader is cautioned that there is an inherent bias between the concentration values reported for these two groups of compounds. If all the data were censored at 0.05 mg/L, then two-thirds (66.4%) of the detections by the GC/MS method would have been reported as less than values (see footnote 4 in Table 2 for the effect of censoring on the most frequently detected pesticides). In some cases, concentrations were quantified above the method detection limit but below the reporting limits reported previously. These concentrations were used for load and yield calculations.

Calculations Pesticide concentrations are not corrected for analytical recovery. Total herbicide (or total insecticide) concentrations are the summed concentrations of all target herbicides (or insecticides) in a sample. For load and yield calculations, concentrations of herbicides (or insecticides) reported as not detected were assigned a value of zero. During periods when no samples were collected, daily pesticide concentrations were obtained

Fig. 1. The distribution of the number of herbicides, insecticides, and total pesticides detected in each sample.

by linear interpolation [9]. A daily pesticide load was calculated (daily concentration times daily discharge) even though samples were not collected every day. The annual load was calculated as the sum of the daily products of concentration and streamflow for a one-year period. The annual yield for a given compound was calculated by dividing its annual load by the urban (or agricultural) land-use area. Land use was delineated using geographic information system [20,21]. Urban land-use area, used in calculating pesticide yields, refers to that part of the total basin area that has urban land use. For all but one site (Norwalk, CT), the urban land-use area represents more than 75% of the total area of the basin, and, in the majority of the remaining sites, urban land-use area represents more than 90% of the total area of the basin (Table 1). RESULTS AND DISCUSSION

Frequency of detection and concentrations of pesticides in urban streams Eight urban streams in the United States were monitored for pesticides and pesticide transformation products (Table 2). The target pesticides consisted of 23 insecticides, 13 of which were detected at least once in one or more of the eight urban streams, and 52 herbicides, 28 of which were detected in one or more of the eight urban streams. Only one of the seven selected transformation products (deethylatrazine) was detected. Of the 215 samples from these streams, only 14 had no pesticide detections (6%), only 17 had no herbicide detections (8%), and only 35 had no insecticide detections (16%). There were two or more herbicides and insecticides detected in 85 and 54% of the samples, respectively. Figure 1 shows the distribution of the number of herbicides, insecticides, and total pesticides detected in each sample. At the national scale, the insecticides most frequently detected in samples from the urban sites were diazinon, carbaryl, chlorpyrifos, and malathion (70, 44, 18, and 14% of samples, respectively; Table 2). These four insecticides all have substantial use in both urban and agricultural areas. Diazinon and carbaryl are used to control insects on turfgrass and gardens in urban areas. Chlorpyrifos and malathion are commonly used on ornamentals and in pet shampoos. Chlorpyrifos also is used to control termites in homes and buildings. Malathion also is used in broad-scale aerial applications to control fruit flies and mosquitoes in urban areas. At the national scale, the herbicides most frequently de-

Comparison of pesticides in eight urban streams

tected in samples from urban streams were simazine, prometon, atrazine, tebuthiuron, and metolachlor (72, 71, 54, 22, and 20% of samples, respectively; Table 2). Prometon, simazine, and tebuthiuron have substantial urban use for weed control on roadways and railways, along fences, and in other public areas. Simazine was used on turfgrass and for control of aquatic weeds. (Simazine is no longer registered for the control of aquatic weeds.) Atrazine has limited nonagricultural use. Metolachlor is restricted to agricultural applications only. Many of the herbicides were detected frequently at only one or a few urban streams. For example, tebuthiuron was detected in 71 and 76% of the samples from Portland, Oregon, and Atlanta, Georgia, respectively, but in only 22% of samples from all eight urban streams combined. Other herbicides detected more frequently in the stream in Portland, Oregon, than in the other urban streams sampled include dichlobenil (18% compared to 1.4% nationally), diuron (36% compared to 10% nationally), and triclopyr (14% compared to 2.5% nationally). Both dichlobenil and diuron are used in public green areas, such as parks, and triclopyr is used on right of ways, industrial sites, and turfgrass. Diuron was detected in 38% of the samples in Las Vegas, Nevada, compared to 10% nationally. In the Las Vegas, Nevada, watershed, the source of diuron may have been runoff from a golf course that was near the sampling site. Herbicides detected more frequently in the stream in Washington, DC, than in the other urban streams sampled include 2,4-D (41% compared to 13% nationally), oryzalin (16% compared to 3.5% nationally), and pendimethalin (39% compared to 11% nationally). Oryzalin, 2,4-D, and pendimethalin are all used to control weeds in turfgrass. In most samples, a few herbicides accounted for most of the total herbicide concentration. As an example, simazine and prometon generally constituted the greatest proportion of the herbicides in samples from Washington, DC. These two herbicides accounted for .40% of the total herbicide concentration on 73% of the sampling dates and .80% of the total herbicide concentration on 36% of the sampling dates. The MCPA and 2,4-D also contributed substantially to the total concentration at certain times of the year. Herbicides used only in agriculture also were observed especially between April and August. This is significant because there is no agricultural land within this watershed. The same general observations can be made for the insecticides in the stream in Washington, DC. The total insecticide concentration was dominated by diazinon and carbaryl. These two insecticides accounted for .80% of the total insecticide concentration on 76% of the sampling dates. The maximum concentrations of individual herbicides ranged from less than the detection limit to 8.2 mg/L in the eight urban streams monitored in this study. The maximum concentrations of individual insecticides ranged from less than the detection limit to 3.2 mg/L (Table 2). When the concentrations of herbicides and insecticides are summed, a wide range of total concentrations is observed. Both the total herbicide and total insecticide concentrations had ranges of almost four orders of magnitude (total herbicides from ,0.01 to 9.9 mg/L and total insecticides from ,0.01 to 6.1 mg/L). Most of the urban stream samples had higher total herbicide concentrations than total insecticide concentrations. The median of the ratio of total insecticides to total herbicides in individual samples was 0.32. In 20% of the urban samples, however, total insecticide concentrations were higher than total herbicide concentrations. This is in contrast to the small agricultural streams

Environ. Toxicol. Chem. 19, 2000

2253

sampled in the U.S. Geological Survey’s National Water-Quality Assessment Program study, where ,2% of the samples from 37 streams had total insecticide concentrations greater than total herbicide concentrations [10]. These findings are consistent with national pesticide use data that show that insecticide use accounts for a greater proportion of total pesticide use in urban areas than in agricultural areas [1].

Occurrence of agricultural pesticides in urban streams Three agricultural pesticides (alachlor, metolachlor, and metribuzin) were detected routinely in the urban streams, and many other agricultural pesticides were detected in isolated instances (Table 2). (In this paper, the term ‘‘agricultural pesticides’’ is defined as those pesticides that are currently registered only for agricultural uses.) The occurrence of agricultural pesticides in urban watersheds may be the result of one or more of the following mechanisms: surface-water transport from agricultural areas upstream of the urban sampling site, atmospheric deposition of the volatilized pesticides from nearby or distant agricultural areas, input from shallow groundwater originating in agricultural areas, and/or unregistered (illegal) use of these compounds within the urban watershed. For urban streams that have a portion of their upstream watershed in agricultural areas, surface-water transport would very likely contribute agricultural pesticides to the urban stream. It is also possible that some of these agricultural pesticides could be introduced to the urban surface-water system by inputs from shallow groundwater recharged in agricultural areas. Data from studies of groundwater in these same urban areas, however, indicate that agricultural pesticides were detected only at low concentrations or not at all [22]. Thus, substantial contributions by this mechanism are unlikely. Another important mechanism by which these pesticides may enter urban streams is atmospheric deposition [23]. Capel et al. [24] measured pesticide concentrations in rain in Minnesota between 1989 and 1994. Herbicides most frequently detected in rain in urban Minneapolis included alachlor, atrazine, metolachlor, and metribuzin. Finally, the unregistered use of agricultural pesticides in the urban areas is possible, but unlikely, and is impossible to quantify. The best evidence that some agricultural pesticides were atmospherically transported from agricultural regions to the urban streams comes from the four watersheds with no agricultural land (Washington, DC; Tallahassee, FL; Atlanta, GA; and Las Vegas, NV). Two additional urban watersheds (Denver, CO, and Portland, OR) contain ,2% agricultural land, suggesting that agricultural land is not a major source of pesticides to these urban streams. The uses of the target pesticides and the urban streams in which they were observed are shown in Table 2. Metolachlor was detected in all six urban streams that drain basins with #2% agricultural land (Table 1). Alachlor and metribuzin were detected in three and four of these streams, respectively (Table 2). It seems reasonable to conclude that these high-use agricultural herbicides did not originate within the urban watersheds. Assuming that there was no unregistered use of these pesticides in these watersheds, the most plausible explanation is that they were transported in the atmosphere from agricultural regions and deposited in the urban basins. Other agricultural pesticides were detected at much lower frequencies in urban streams (Table 2). The herbicides cyanazine, S-ethyl dipropylthiocarbamate, molinate, napropamide, propanil, terbacil, and triallate and the insecticides parathion

Environ. Toxicol. Chem. 19, 2000

6.4 13 19 3.8 3.5 17 24 67 0.059 2.8 8.3 0.037 , 6.8 1.8 4.3 9.6 8.4 130 , 15 13 4.5 74 0.51 16 28 , 1.7 8.4 , 51 , 8.2 2.5 0.072 , 25 , 0.19

0.19 150 11 7.4 5.7 30 28 140 0.91 1.5 0.59 0.23 0.38 7.4 1.4 12 2.0 15 8.9 0.18 3.9 11 51 5.7

, , , 17 , 43 , 6.2

3.9 3.5 20 15 19 16 0.61 53

Diazinon Chlorpyrifos Carbaryl 2,4-D Tebuthiuron Simazine Prometon Metolachlor Diuron

Albany, NY Atlanta, GA Denver, CO Las Vegas, NV Norwalk, CT Portland, OR Tallahassee, FL Washington, DC

Geographic patterns of pesticide occurrence in urban streams Some common seasonal patterns in pesticide concentrations were apparent among streams in similar geographic/climatic

Table 3. Annual yields (g/km2/year) of the most commonly detected pesticides for the eight urban watersheds

Pesticide yields in urban streams Estimates of the mass of pesticides applied in urban areas are not available. Thus, the pesticide loads observed in urban streams can not be compared with the amounts of pesticides applied in the drainage basins, as has been done for agricultural pesticides [9,10]. Comparisons of pesticide loads in various urban streams can be made, however, by using annual pesticide yields. The annual pesticide yield is the mass of pesticides transported in an urban stream in a one-year period divided by the urbanized area of the basin. Table 1 summarizes the total herbicide and total insecticide yields for the eight urban streams. In the eight streams discussed in this paper, the median herbicide yield is 81 g/km2/year and the median insecticide yield is 31 g/km2/year. Both the maximum herbicide yield (320 g/km2/year) and maximum insecticide yield (170 g/km2/year) were observed in Washington, DC. The stream in Denver, Colorado, had a total insecticide yield (160 g/km2/year) comparable to that of the Washington, DC, stream. The annual yields of individual pesticides varied widely among the eight urban streams (Table 3). Yields of several pesticides were higher in Washington, DC, than in the other urban streams, including the herbicides 2,4-D, metolachlor, and prometon and the insecticides diazinon, fonofos, and malathion. Yields of the herbicides diuron and tebuthiuron and the insecticide carbofuran were highest in Portland, Oregon. The widely used insecticides carbaryl and chlorpyrifos had their highest yields in Denver, Colorado. Atrazine yields in Tallahassee, Florida, and Atlanta, Georgia, were higher than in the other six urban streams studied. The annual yields of the pesticides from urban basins were compared to the yields from nearby agricultural basins for six of the eight urban sites (Table 1). For agricultural basins, the annual pesticide yield is the mass of pesticides transported in a stream in a one-year period divided by the agricultural area of the basin. The median herbicide yield from the urban basins (80 g/km2/year) was about four times smaller than the median herbicide yield from the agricultural basins (300 g/km2/year). The median urban insecticide yield (31 g/km2/year), however, was about five times greater that the median insecticide yield from the agricultural basins (6 g/km2/year). In the six sets of paired urban/agricultural basins, the urban basins had a greater yield of herbicides in two of the six basins and a greater yield of insecticides in five of the six basins. Insecticides contributed only 1 to 11% (median 3%) of the total pesticide yield from the agricultural basins, but they contributed 12 to 67% (median 29%) of the total pesticide yield from the urban basins.

Malathion

and propargite were detected in at least one of the streams with #2% agricultural land in their drainage basins. The occurrence of these pesticides in urban streams is probably also the result of atmospheric transport from areas where they are used in agriculture. Capel et al. [24] detected many of these same pesticides in precipitation in Minneapolis and suggested that urban areas may be excellent locations for monitoring regional use and changes in use for many agricultural pesticides. They suggested that deposition in an urban area can be representative of the regional background concentrations of pesticides since the urban atmosphere is removed from direct agricultural sources.

, 0.68 4.9 2.3 0.19 2.1 0.58 14

R.S. Hoffman et al.

Atrazine

2254

Comparison of pesticides in eight urban streams

Fig. 2. Temporal concentration patterns of total herbicides (black lines) and total insecticides (dashed lines) in USA streams near (A) Washington, DC; (B) Norwalk, Connecticut; (C) Albany, New York; (D) Atlanta, Georgia; and (E) Tallahassee, Florida. Concentration scales vary among graphs.

areas. Of the eight urban streams in this study, three (Norwalk, CT; Washington, DC; and Albany, NY) were in the northeastern United States, two (Atlanta, GA, and Tallahassee, FL) were in the southeastern United States, and three (Denver, CO; Las Vegas, NV; and Portland, OR) were in the western United States. Some generalizations regarding concentration patterns observed in the northeastern and southeastern regions can be made. In general, the temporal concentration patterns in the urban streams were consistent with the characteristics of the local growing season, with the exception of arid Las Vegas, Nevada, which did not have any observable seasonal pattern, as it does not have a distinct growing season. The three urban streams in the northeastern United States had similar seasonal patterns with respect to total herbicide and total insecticide concentrations (Figs. 2A to C). Total herbicide concentrations were highest in the spring and early summer and then decreased to minimal levels during the winter months. Total insecticide concentrations generally were highest in midsummer and early autumn in the three northeastern streams (Figs. 2A to C). Total insecticide concentrations were very low or below detection levels between December and

Environ. Toxicol. Chem. 19, 2000

2255

March in all three streams in the northeastern United States. The periods of elevated total herbicide and total insecticide concentrations in these streams coincide with the turfgrass and garden growing season and the most active season for insects, respectively [25]. The temporal concentration pattern in these streams reflects the timing of pesticide application for weed and insect control. The pattern of pesticide concentrations in Denver, Colorado, was similar to the pattern observed in the northeastern United States. Elevated concentrations were observed during the active growing season [25] from May to September, followed by relatively constant, low levels for the rest of the year, although there were no insecticides detected in November and December. The occurrence of both herbicides and insecticides in the Denver, Colorado, stream for most of the year reflects the relatively short dormant season in this area. Seasonal concentration patterns of total herbicides and total insecticides in the two urban streams in the southeastern United States were different than the pattern in the northeastern region (Figs. 2D to E). In Atlanta, Georgia (Fig. 2D), the highest total herbicide concentrations occurred in the winter (November to January) and early spring (March and April). During the remaining months, herbicide concentrations were much lower but consistently above the detection limit. In both streams in the southeastern United States, the total insecticide concentrations varied throughout the year, reaching their maxima during the late spring and summer, but had no discernible overall pattern (Figs. 2D to E). Total insecticide concentrations were relatively high during the winter months. The growing season for both weeds and insects in this region is much longer than in the northeastern United States as a result of the warmer climate. The flow in the urban stream in Las Vegas, Nevada, consists almost entirely of sewage treatment plant effluent and urban runoff and is relatively constant except after storm events. Concentrations of both herbicides and insecticides were relatively constant throughout the year except for a few spikes. These concentration spikes always coincided with a sharp rise in stream discharge due to storm runoff. One such event occurred during December, resulting in elevated concentrations of the herbicides diuron and prometon and the insecticides diazinon and malathion. In general, both herbicides and insecticides were detected throughout the year, which is consistent with the fact that this area has no substantial dormant season [25]. The period of lowest concentrations for both herbicides and insecticides (August through October) coincided with the period of low discharge.

Factors affecting pesticide occurrence in urban streams Numerous factors may contribute to differences in pesticide occurrence (as quantified by frequency of detection, load, and yield) in these eight urban streams. Regional use of specific herbicides and insecticides to control particular pests is the most obvious factor. This phenomenon is well known anecdotally but cannot be quantified because sales data are not available. Various other factors that could be quantified were considered as possible causative factors, including annual precipitation, population density (Table 1), and median and per capita income. Only herbicides and insecticides that have substantial use in urban areas were considered in this analysis. Atrazine was not included even though it has limited urban use. Of the factors listed previously, population density was the

2256

Environ. Toxicol. Chem. 19, 2000

R.S. Hoffman et al.

Fig. 3. Relation of population density and (A) total herbicide yield and (B) total insecticide yield in eight urban streams based on area of urbanized land. Only pesticides with urban use were included; 2,000 persons/km2 is typical of established residential area of a major city with single-family homes, 1,000 persons/km2 is typical of an established suburb with lots ;0.5 acre, and 200 persons/km2 is typical of a currently developing suburb with lots ;5 acres. The lines are based on a polynomial regression but are presented only as a visual aid.

only one that showed a substantial relation with pesticide yields. Population density in urban areas may influence pesticide occurrence in streams as follows. Pesticide yield would be expected to be relatively low in urban areas of both low and high population density. Low population densities in urban or suburban areas translate to large lot sizes that would likely have only a small portion of turfgrass and gardens potentially controlled for pests. High population densities in urban or suburban areas translate to small lot sizes that would likely have little to no green area that is chemically treated. Pesticide yield is most likely to be highest in urban or suburban areas of moderate population density, where lot sizes have relatively large proportions of turfgrass and gardens potentially controlled for pests. The findings of this study are in general agreement with this hypothesis for both total herbicide yield

(Fig. 3A) and total insecticide yield (Fig. 3B). However, further studies with more data and better controls would be needed to make definitive conclusions, especially from watersheds that have population densities in the range of 1,000 to 2,000 people/ km2.

Pesticide concentrations and ecosystem concerns At some concentration, most pesticides will begin to have an adverse effect on surface-water ecosystems. Although many different pesticides were detected in the urban streams in this study, their impact is largely unknown and is a difficult issue to address. One way to address the issue is to compare the observed concentrations with criteria that have been established to protect the health of aquatic organisms. Criteria for the protection of aquatic organism health have been established

Fig. 4. Seasonal patterns of selected pesticide concentrations in Cherry Creek, Denver, Colorado, USA, normalized to their aquatic-life criteria value. The water-quality criteria values used in the normalization are given in the legend.

Comparison of pesticides in eight urban streams

for only 28 of the 82 pesticides and transformation products targeted in this study (Table 2). These criteria have been established by the U.S. Environmental Protection Agency [26], Environment Canada [27] (http://www.ec.gc.ca/ceqg-rcqe, accessed on March 22, 2000), the Canadian Council of Resource and Environment Ministers [28], and the Great Lakes International Joint Commission [29]. The concentrations of only five pesticides exceeded an aquatic-life criterion in one or more samples in this study. All five were insecticides (carbaryl, chlorpyrifos, diazinon, malathion, and parathion). Most notably, carbaryl exceeded its criterion in 10% of samples, and diazinon exceeded its criterion in 17% of samples from the eight streams. The remaining insecticides exceeded their criteria in only one or a few samples. One way to compare the ambient concentrations to the criteria is illustrated in Figure 4 for samples from Denver, Colorado. A value of one in this plot means that the concentration of the pesticide in the stream was equal to its criteria concentration. Points above the line represent concentrations of an individual pesticide that exceeded its criteria value. Five insecticides in this stream exceeded their criterion in at least one sample. In some samples, more than one insecticide exceeded a criterion. In Denver, as well as the other urban sites, insecticides exceeded criteria concentrations most often in the spring and summer. This coincides with the active season for most insects and, thus, the period of greatest insecticide application [25]. This is also the period of the year when reproduction and early growth of most aquatic organisms occurs, which could have implications for ecosystem health. The co-occurrence of multiple pesticides in the urban streams was commonly observed. In 46% of the samples, more than one insecticide was observed (Fig. 1). Usually, there was some combination of carbaryl, chlorpyrifos, diazinon, and/or malathion. All four of these compounds are cholinesterase inhibitors; the latter three are organphosphorous compounds. Bailey et al. [30] showed that chlorpyrifos and diazinon exhibit additive toxicity when present together in both laboratory and field tests. For herbicides, the co-occurrence of multiple compounds was even more common. Four or more herbicides were quantified in 61% of the water samples obtained from the eight urban streams. For most of the time, each of the individual herbicides and insecticides were below its water-quality criterion for the protection of aquatic health (if it has one), but the additive and synergistic effects of multiple herbicides and insecticides on the aquatic biota are largely unknown.

Importance of urban areas as a source of pesticides to surface waters Comparisons of the estimated mass of pesticides exported from urban and agricultural areas can be used to help assess the overall importance of urban areas as a source of pesticides to the nation’s surface water. The following estimates are based on data from only eight urban watersheds and 32 small agricultural watersheds [10]; thus, the uncertainty in the calculated values may be large. Nonetheless, even order-of-magnitude comparisons are useful in assessing the role of urban areas as a source of pesticides. The total national export of urban pesticides to surface waters is estimated by multiplying the median, minimum, and maximum annual yields among the eight urban watersheds (Table 1) by the area of urbanized land in the conterminous United States (260,000 km2; [31]). Similarly, the total national export of agricultural pesticides to surface

Environ. Toxicol. Chem. 19, 2000

2257

waters is estimated by multiplying the median, minimum, and maximum annual yields among the 32 watersheds [10] by the area of row crops, orchards, vineyards, and nurseries in the conterminous United States (1,050,000 km2; [20]). On the basis of median yields, the annual export of herbicides from agricultural areas is 110 metric tons (range 1.1– 2,500 metric tons), whereas the annual export of herbicides from urban areas is only 6 metric tons (range 2–80 metric tons). Comparing the values based on the median yields, the mass of herbicides contributed to surface waters by agricultural areas is 20 times the mass contributed by urban areas. When the same calculation is done for insecticides, a different result is observed. The annual export of insecticides from agricultural areas is 4 metric tons (range 0.05–280 metric tons), whereas the annual export of insecticides from urban areas is 8 metric tons (range 2–40 metric tons). Based on the median yield values, the contribution of insecticides from agricultural areas is similar to the contribution from urban areas. SUMMARY

This is the first large-scale examination of pesticides in urban surface waters in the United States since the early 1980s and the only large-scale study that has analyzed for the current high-use pesticides. Although there were differences among the eight urban areas, there were some important commonalties. Most of the samples contained at least one herbicide (97%) or one insecticide (89%). The seasonal concentration patterns that were observed were consistent with the local growing and dormant seasons. At many sites, some of the pesticides detected in the urban streams were registered only for agricultural use. Their presence in the urban streams was possibly due to atmospheric deposition. Perhaps the most unexpected outcome of this study is the importance of the insecticides as quantified by the yield of urban watersheds and by their concentrations in the urban streams in relation to aquatic health criteria. The results of this study suggest that not enough is known about pesticides in urban streams and that urban areas should not be overlooked when assessing the sources and monitoring the occurrence of pesticides in surface waters. Acknowledgement—This research was conducted at part of the U.S. Geological Survey National Water-Quality Assessment Program’s National Pesticide Synthesis Project. R.S. Hoffman was partially supported by the Gibson Hydrogeology Endowment through the University of Minnesota Department of Geology and Geophysics. REFERENCES 1. Aspelin AL. 1998. Pesticides industry sales and usage: 1994 and 1995 market estimates. EPA 733-R-97-002. U.S. Environmental Protection Agency, Washington, DC. 2. Barbash JE, Resek EA. 1996. Pesticides in Groundwater: Distribution, Trends, and Governing Factors. Lewis, Chelsea, MI, USA. 3. Gianessi LP, Anderson JE. 1996. Pesticide use in U. S. crop production. National Data Report. National Center for Food and Agricultural Policy, Washington, DC. 4. Hodge JE. 1993. Pesticide trends in the professional and consumer markets. In Racke KD, Leslie AR, eds, Pesticides in Urban Environments: Fate and Significance. ACS Symposium Series 522. American Chemical Society, Washington, DC, pp 11–17. 5. Rumbaker RV, Matter RM, Clement DP, Erickson FK. 1972. Use of pesticides in suburban homes and gardens and their impact on the aquatic environment. EPA 733-R-97-002. U.S. Environmental Protection Agency, Washington, DC. 6. Whitmore RW, Kelly JE, Reading PL. 1992. Executive summary, results and recommendations: Vol 1. National home and garden

2258

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

Environ. Toxicol. Chem. 19, 2000

pesticide use survey. EPA RTI/5100/17-01F68-WO-0032. U.S. Environmental Protection Agency, Research Triangle Park, NC. Immerman FW, Drummond DJ. 1985. National urban pesticide applicator survey: Overview and results. Final Report. RTI 2764/ 08–01F. Research Triangle Institute, Research Triangle Park, NC, USA. Goolsby DA, Thurman EM, Kolpin DW. 1991. Geographic and temporal distribution of herbicides in surface waters of the upper midwestern United States,1989–1990. USGS WRIR 94-4015. U.S. Geological Survey, Reston, VA, pp 183–188. Larson SJ, Capel PD, Goolsby DA, Zaugg SD, Sandstrom MW. 1995. Relations between pesticide use and riverine flux in the Mississippi River Basin. Chemosphere 31:3305–3321. Larson SJ, Capel PD, Gilliom RJ. 1999. Pesticides in streams of the United States: Initial results from the National Water-Quality Assessment Program. USGS WRIR 98-4222. U.S. Geological Survey, Sacramento, CA. Struger J, Boyter D, Licsko ZJ, Johnson BD. 1995. Environmental concentrations of urban pesticides. Proceedings, 38th Conference of the International Association of Great Lakes Research, East Lansing, MI, USA, p 127. Cole RH, Frederick RE, Healy RP, Rolan RG. 1984. Preliminary findings of the priority pollutant monitoring project of the nationwide urban runoff program. J Water Pollut Control Fed 56: 898–908. Oltmann RN, Shulters MV. 1989. Rainfall and runoff quantity and quality characteristics in Fresno, California, October 1981 to April 1983. USGS WSP 2335. U.S. Geological Survey, Sacramento, CA. Wotzka PJ, Lee J, Capel PD, Lin M. 1994. Pesticide concentrations and fluxes in an urban watershed. AWRA Technical Publication Series TPS-94-4. American Water Resources Association, Herndon, VA, pp 135–145. Harrison SA, Watschke TL, Mumma RO, Jarrett AR, Hamilton GW Jr. 1993. Nutrient and pesticide concentrations in water from chemically treated turfgrass. In Racke KD, Leslie AR, eds, Pesticides in Urban Environments: Fate and Significance. ACS Symposium Series 522. American Chemical Society, Washington, DC, pp 191–207. Gilliom RJ, Aley WM, Gurtz ME. 1995. Design of the national water-quality assessment program: Occurrence and distribution of water-quality constituents. USGS Circular 1112. U.S. Geological Survey, Sacramento, CA. Shelton LR. 1994. Field guide for collecting and processing stream-water samples for the national water-quality assessment program. USGS OFR 94-455. U.S. Geological Survey, Sacramento, CA. Zaugg SD, Sandstrom MW, Smith SG, Fehlberg KM. 1995. Methods of analysis by the United States Geological Survey National Water-Quality Laboratory: Determination of pesticides in water by C-18 solid-phase extraction and capillary-column gas chro-

R.S. Hoffman et al.

19.

20. 21. 22.

23. 24. 25. 26. 27. 28.

29.

30. 31. 32. 33.

matography/mass spectrometry with selected-ion monitoring. USGS OFR 95-181. U.S. Geological Survey, Denver, CO. Werner SL, Burkhardt MR, DeRusseau SN. 1996. Methods of analysis by the United States Geological Survey National Water Quality Laboratory: Determination of pesticides in water by Carbopak-B solid-phase extraction and high-performance liquid chromatography. USGS OFR 96-216. U.S. Geological Survey, Denver, CO. Gilliom RJ, Thelin, GP. 1997. Classification and mapping of agricultural land for the national water-quality assessment. USGS Circular 1131. U.S. Geological Survey, Sacramento, CA. Hitt KJ. 1994. Refining 1970s land-use data with 1990 population data to indicate new residential development. USGS WRI 944250. U.S. Geological Survey, Reston, VA. Kolpin DW, Barbash JE, Gilliom RJ. 1998. Occurrence of pesticides and in shallow groundwater of the United States: Initial results from the National Water-Quality Assessment Program. Environ Sci Technol 32:558–566. Majewski MS, Capel PD. 1995. Pesticides in the Atmosphere: Distribution, Trends, and Governing Factors. Lewis, Chelsea, MI, USA. Capel PD, Ma L, Wotzka PJ. 1998. Wet deposition of pesticides in Minnesota, 1989–1994. USGS WRIR 96-3165. U.S. Geological Survey, Mound View, MN. Bowker M. 1991. Caring for Lawns. NK Lawn and Garden Co., Corte Madera, CA, USA. U.S. Environmental Protection Agency. 1999. National recommended water quality criteria. USEPA 822-Z-99-001. Washington, DC. Environment Canada. 1999. Canadian water quality guidelines for the protection of aquatic life. Ottawa, ON. Canadian Council of Ministers of the Environment. 1998. Canadian sediment quality guidelines for the protection of aquatic life: Introduction and summary tables. CCME EPC-98E. Canadian Council of Ministers of the Environment, Winnipeg, MB. International Joint Commission Canada and United States. 1987. Revised Great Lakes water quality agreement of 1978—Agreement, with annexes and terms of reference, between the United States and Canada signed at Ottawa November 22, 1978 and phosphorus load reduction supplement signed October 16, 1983 as amended by protocol signed November 18, 1987. International Joint Commission Canada and United States, Windsor, ON, Canada. Bailey HC, Miller JL, Miller MJ, Wiborg LC, Deanovic L, Shed T. 1997. Joint acute toxicity of diazinon and chlorpyrifos to Ceriodaphnia dubia. Environ Toxicol Chem 16:2304–2308. World Resources Institute. 1996. World Resources, 1996–1997: A Guide to the Global Environments—The Urban Environment. Oxford University Press, Oxford, UK. Smith MD. 1996. The Ortho Problem Solver, 4th ed. Soloris Group, San Ramon, CA, USA. Meister RT, ed. 1998. Farm Chemicals Handbook ’98. Meister, Willoughby, OH, USA.