Organophosphorus Insecticides in Agricultural and Residential Runoff

Feb 22, 2006 - University of California, Los Angeles, California 90095. Development of total ... irrigation and storm runoff events at agricultural si...
0 downloads 0 Views 489KB Size
Environ. Sci. Technol. 2006, 40, 2120-2127

Organophosphorus Insecticides in Agricultural and Residential Runoff: Field Observations and Implications for Total Maximum Daily Load Development JOEL A. PEDERSEN,‡ MATT A. YEAGER,§ AND I . H . ( M E L ) S U F F E T * ,† Environmental Science and Engineering Program, University of California, Los Angeles, California 90095

Development of total maximum daily loads (TMDLs) for nonpoint source pollutants requires mass flux estimates for targeted compounds from contributing sources. We measured organophosphorus insecticide concentrations in surface runoff from agricultural and residential land-use sites in a southern Californian watershed over the course of runoff-producing irrigation and rainfall events. Event mean concentrations (EMCs) for chlorpyrifos, diazinon, and malathion exhibited considerable variability among irrigation and storm runoff events at agricultural sites; residential storm runoff EMCs for these compounds were considerably less variable. Event loads and EMCs were higher for runoff events following reported insecticide applications. Organophosphorus insecticide EMCs were not consistently correlated with hydrologic characteristics of runoff events. Our results indicate that on an area basis, loads from residential land may exceed those from sites planted in row crops for a given rainfall depth, suggesting that residential land use warrants explicit consideration in TMDL development and implementation. No consistent first flush effect was discernible for organophosphorus insecticides in storm or irrigation runoff. A relative potency factor approach is introduced to permit evaluation of organophosphorus insecticides on a common toxicological basis and allow development of TMDLs and pollutant control strategies for these compounds as a class.

Introduction Organophosphorus insecticides rank among the most widely used insect control agents in the United States. Their moderate persistence coupled with frequent application in intensively cultivated areas can result in persistent surface water contamination. Concentrations in stream systems can cause toxicity to aquatic organisms (1, 2). Surface runoff from treated lands represents an important pathway for unintentional introduction of these compounds into aquatic ecosystems. Runoff-producing irrigation and precipitation * Corresponding author phone: (310)206-8230; fax: (310)206-3358; e-mail: [email protected]. † University of California. ‡ Present address: Department of Soil Science, University of Wisconsin, Madison, WI 53706; [email protected]. § Present address: San Bernardino County Flood Control District, San Bernardino, CA 92415; [email protected]. 2120

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 7, 2006

events mobilize and transport dissolved and suspended sediment-associated pesticides into adjacent water bodies, where they may inadvertently impact nontarget aquatic invertebrates and fish. The ubiquitous occurrence of organophosphorus insecticides in U.S. streams and their presence in surface waters draining agricultural and urban areas have been documented (e.g., 1, 3-5). Most previous studies attempting to attribute receiving water concentrations of organophosphorus insecticides to specific land uses investigated the occurrence of these compounds in streams draining catchments dominated by single land uses (e.g., 1, 6-8); organophosphorus insecticide concentrations were higher in streams receiving runoff from residential areas than those draining predominately commercial and industrial land uses (8). Few studies have examined organophosphorus insecticide inputs from multiple land uses within a single watershed. Most studies making comparisons among land uses measured concentrations in streams draining catchments with one predominant land use (9-11); dilution with nonrunoff baseflow and runoff from secondary land uses during storm events can complicate attribution of inputs to specific sources in this approach (11). Determination of organophosphorus insecticide concentrations and mass flux in surface runoff from well-defined areas of single land uses (e.g., edge-of-field measurements) is essential for developing and evaluating management strategies to minimize their offsite movement and protect sensitive aquatic organisms. Schiff and Sutula (5) recently examined chlorpyrifos and diazinon concentrations in stormwater runoff from homogeneous land-use catchments for a limited number of rainfall events and found diazinon present in nearly all samples. Chlorpyrifos and diazinon loads have been reported for streams receiving storm and irrigation runoff from agricultural areas in the Central Valley, California (e.g., 12-16). The occurrence of pollutants that impair the beneficial uses of surface waters triggers the requirement for total maximum daily load (TMDL) development under the U.S. Clean Water Act §303(d). A TMDL entails calculation of the maximum amount of a pollutant a water body can assimilate and still maintain its beneficial uses and allocation of that amount to the pollutant’s sources. Quantification of mass loads and characterization of temporal variation in loadings would greatly improve source allocation of nonpoint source pollutant loads. Numerous streams in California, and those in several other states, require TMDL development for organophosphorus insecticides, particularly chlorpyrifos and diazinon (17). TMDL target levels and pollutant control strategies are typically developed for individual contaminants. However, focus on individual chemicals may not lead to remedies protective of aquatic life when mixtures of chemicals sharing a mechanism of toxicity are present. Organophosphorus insecticides exert their toxicity by inhibiting the enzyme acetylcholinesterase. Because these compounds share a primary mode of action and are often present in the environment as mixtures, consideration of their combined toxicity appears warranted when assessing potential ecological effects and developing pollution control strategies. Organophosphorus insecticides can be assessed as a class using a relative potency factor approach (18, 19). The primary objective of this study was to characterize and quantify organophosphorus insecticide inputs from residential and agricultural land uses in a mixed-use watershed to support TMDL development. A secondary objective was to assess the potential ecotoxicological significance of 10.1021/es051677v CCC: $33.50

 2006 American Chemical Society Published on Web 02/22/2006

FIGURE 1. Calleguas Creek watershed (880 km2) land use. Predominant land uses in the watershed were residential (11%; 55% of urban land use), irrigated row crops (9%), and irrigated orchards (8%). Most of the remaining area was undeveloped (60%). The crops represented in the seven sampling sites (A1-A7) accounted for ∼55% of reported organophosphorus insecticide application to irrigated row crops in 1999 in the watershed (25). One lemon orchard site (O1) was selected to represent tree crop land use. Lemon orchards accounted for >97% of reported organophosphorus insecticide application to tree crops in the watershed (25). R1 and R2 are residential land-use sites. Based on 2000 data from the Southern California Association of Governments. their combined presence in runoff using a relative potency factor approach. We measured organophosphorus insecticide concentrations in surface runoff from seven irrigated row crops, one irrigated orchard, and two residential sites during irrigation and rainfall events in a coastal southern California watershed. The bulk of our analysis focused on water phase concentrations because dissolved organophosphorus insecticides are the most immediately bioavailable to in-stream organisms. Our field observations are most directly transferable to irrigated agricultural settings (especially in Mediterranean climates) and suburban housing developments. The implications for TMDL development are national in scope, and the approach taken may be applicable to other pesticide classes.

Materials and Methods Site Description. The Calleguas Creek watershed is located in coastal southern California (Figure 1). Annual precipitation averages 33 cm on the coastal plain, with 85% occurring between November and March. We selected sites representing agricultural and residential land uses in the lower watershed based on uniformity of land use, a defined drainage area, ease of access and flow measurement, and proximity to stream (Table 1). Sample Collection. We sampled dry weather irrigation runoff from crop sites between July and early December 1999. These fields were often irrigated in sections; therefore, sampled events often represented runoff from irrigation of discrete portions of larger fields. Wet weather sampling was conducted at four row crop sites, one orchard site, and two residential sites. Because the bulk of precipitation-induced pesticide runoff often occurs in response to major storm

events (20), we targeted storms predicted to yield g15 mm of rainfall. Runoff samples (2 L) were collected at 15- to 60-min intervals over the course of selected runoff-producing events by either grab sampling (irrigation runoff, storm runoff from orchard and residential sites) or by autosampler (storm runoff from row crop sites). For the majority of events, flow measurements from individual fields were made using an ultrasonic area-velocity meter. In some instances, flow was estimated from manual water level and velocity measurements. Additional sample collection and flow measurement details are provided in the Supporting Information. During each sampling event, one or more stream water samples were collected from Conejo Creek, a tributary to Calleguas Creek. These samples were not used for quantitative load estimates but to make comparisons with runoff samples using the relative potency factor approach introduced below. Sample Extraction. Suspended sediment was separated from the water phase by filtration through 0.7-µm glass fiber filters. Dissolved and sorbed phases were thus operationally defined as those passing or retained by the 0.7-µm filter. Aqueous samples were liquid-liquid extracted with dichloromethane (21). Solid-phase samples were dried, and a subset was extracted with supercritical CO2 (22). The Supporting Information further details the extraction methods. Instrumental Analysis. Sample extracts were analyzed for chlorpyrifos, diazinon, dichlorvos, dimethoate, disulfoton, and malathion by gas chromatography with flame photometric detection (23) (see the Supporting Information for details). Data Analysis. We calculated several pollutant load metrics to facilitate comparison of runoff events: event mass load (M), event mean concentration (EMC), and a first flush VOL. 40, NO. 7, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2121

TABLE 1. Sampling Site Characteristicsa reported organophosphorus insecticide contributing area (ha) siteb

dw

A1 A2 A3 A4g A5 A6g A7g O1 R1 R2

1.3 2.7 2.6-4.9 2.5 4.6 2.4 2.0-2.7 ns ns ns

crop

wwc 18.7 9.9 45.1 ns 4.7 ns ns 95.4 0.1 0.5

applications (kg)d

events

land use irrigated crop irrigated crop irrigated crop irrigated crop irrigated crop irrigated crop irrigated crop orchard residential residential

dw green onion strawberry corn cucumber zucchini/cilantro pepper celery lemon

sampled ww greense

various kale spinach/cilantro ns cilantro ns ns lemon

dw

ww

1 1 4 1 2 1 4 0 0 0

2 3 4 0 3 0 0 2 3 3

dw chl

dzn

ww dim

chl

dzn

dim

14.9

1.4

ns

ns

1.8 ns

ns ns 26.4

ns ns

ns ns

5.1 50.7f

10.5f

a Abbreviations: chl, chlorpyrifos; dim, dimethoate; dw, dry weather; dzn, diazinon; ns, not sampled; ww, wet weather. b Row crop sites were chosen based on the representativeness of the crops. Two sites in a suburban housing tract that received runoff from multiple single-family dwellings were chosen to represent residential land use. c Contributing areas during wet weather events were based on field observations and measured using a global positioning system. d Applications during dry weather (Jul 17, 1999 to Dec 10, 1999) and wet weather (Dec 11, 1999 to Apr 17, 2000) sampling as reported by the California Department of Pesticide Regulation PUR database (25). Table S5 in the Supporting Information provides application dates and amounts per application, including applications prior to the sampling period. The PUR database does not include information about home and garden applications. The mass of nonagricultural applications of chlorpyrifos, diazinon, and malathion by licensed pesticide control operators in Ventura County reported in the PUR database were 0.08, 0.28, and 0.04 of those to agricultural land uses. e Kale, lettuce, Chinese cabbage, bok choi. f Field area varied among irrigation runoff events during dry weather sampling. Insecticide application calculated for 4.9- and 2.7-ha fields for sites A3 and A7. g Site was not sampled for wet weather runoff due to unsafe conditions during storms.

metric (see the Supporting Information for details). For comparison purposes, we grouped metrics into land-useevent-type categories (viz., agricultural irrigation, agricultural storm and residential storm runoff) that we refer to as runoff regimes. The low number of storm runoff events sampled at the orchard site precluded statistical comparisons with other land-use categories. We tested whether the derived metrics for each regime appeared to be drawn from underlying normal or log-normal distributions and employed parametric or nonparametric tests as appropriate (see the Supporting Information for details).

Results and Discussion Although all sampled storms produced runoff from the irrigated row crop sites, some (viz., A4, A6, and A7) were not sampled during storms due to unsafe conditions. Only two sampled storms produced runoff at site O1. Because row crop sites were often irrigated until shortly before rain events, antecedent soil moisture was often high resulting in rapid exceedance of infiltration capacity and early initiation of overland flow. In contrast, microsprinkler irrigation of site O1 increased soil moisture only immediately adjacent to the trees. Runoff occurred at this site fairly late in the rainy season and only during major storms (precipitation g23 mm). Table 2 and Table S2 in the Supporting Information summarize storm characteristics and hydrologic data from the sampled events. Concentration Profiles. Organophosphorus insecticides usually occurred as mixtures and were present in the dissolved phase of runoff from all sites during each event sampled. We screened a subset of suspended sediment samples (usually two per event) for organophosphorus insecticides focusing on samples with high dissolved concentrations or large suspended sediment loads. Detection of organophosphorus insecticides in suspended sediment extracts triggered analysis of additional samples from that event. Few suspended sediment extracts contained detectable organophosphorus insecticides (14 of 63 samples analyzed); these occurred in three events. Runoff from residential land-use sites carried very small suspended sediment loads. We therefore focused our evaluation on dissolved organophosphorus insecticides; suspended sediment-associated compounds are discussed where appropriate. 2122

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 7, 2006

TABLE 2. Characteristics of Sampled Stormsa date Jan 15, 2000 Feb 12, 2000 Feb 23, 2000 Apr 17, 2000

tr P Imax tmax ADWP AP7 AP14 (min) (mm) (mm‚hr-1) (min) (days)b (mm)b (mm)b 375 135 405 405

17 15 23 36

4 12 16 16

195 15 330 150

63 1 2 42

1 12 39 0

5 20 80 0

a Abbreviations: ADWP, antecedent dry weather period; AP , depth 7 of rainfall recorded during antecedent 7 days; AP14, depth of rainfall recorded during antecedent 14 days; Imax, event maximum rainfall intensity; P, total depth of precipitation; tmax, time when Imax recorded from beginning of event; tr, duration of rainfall. Rainfall measurements were taken in Camarillo at the California State University Channel Islands campus. Because only storms capable of producing runoff were of interest, the minimum rainfall depth used in calculating ADWP, AP7, and AP14 was 2.5 mm. A 24-h interevent time was used to define discrete storm events. Sampled storms represented 31% of potential runoffproducing rainfall events and 44% of rainfall volume for the 1999-2000 storm season. b During the rainy season, row crop sites were irrigated between storms. The ADWP, AP7, and AP14 therefore provide information on the temporal proximity and magnitude of preceding storm events but cannot be used to estimate soil moisture conditions for irrigated sites.

Organophosphorus insecticide concentrations varied considerably over the course of each event, sometimes more than 10-fold (Figure 2 and Figure S2 in the Supporting Information). Fluctuations in insecticide concentrations were often synchronous. Maximum concentrations sometimes occurred near the beginning of the event (e.g., malathion in Figure 2a, diazinon in Figure 2b), but this was not a consistent phenomenon (e.g., all compounds in Figure S1a, diazinon in Figure S1b in the Supporting Information). Variation in concentration peak timing among events was likely attributable to differences in compound availability for extraction into overland flow and surface runoff hydrodynamics. Sampling over the entire runoff hydrograph was required to capture the sometimes large concentration variations. Previous investigators have noted dramatic within-event variability in pesticide concentrations (5, 12-15, 24). Event Loads. Dissolved chlorpyrifos, diazinon, and malathion were frequently detected in runoff from both agricultural and residential land uses. Loads of these compounds within runoff regimes varied considerably (Figure

FIGURE 2. Temporal variation in dissolved organophosphorus insecticide concentration for (a) the Oct 15, 1999 irrigation event at site A7 and (b) the April 17, 2000 storm event at site A2. 3a). Comparisons of loads among sites or between irrigation and storm events at the same site were not generally useful because runoff-contributing areas often differed (Table 1). Contributing areas for irrigation events were often smaller than those for storm events at the same sites. Since fields were often irrigated in sections, reported event loads represent insecticide masses contributed by well-defined portions of fields. However, under storm conditions, entire fields contributed runoff. For example, the chlorpyrifos load for the Oct 15, 1999 irrigation event at site A3 (mass, M ) 24 mg) was contributed by a 2.6-ha area over a 14-h period. The entire 45-ha field contributed runoff during storm events (Table 1), yet required several days to be completely irrigated by sections. For individual sites, dissolved loads varied considerably among events for each runoff regime. Loads from individual sites for chlorpyrifos, diazinon, and malathion in agricultural irrigation runoff varied by up to 1.6 orders of magnitude; variability in storm runoff loads from the same sites was sometimes larger. Large variations (3.4 orders of magnitude) in diazinon loads among storm events at site A1 were due to the large load contributed by the April 17, 2000 event (M

) 9.6 g). Storm event loads from individual residential sites ranged over nearly 2 orders of magnitude for chlorpyrifos; less variability was observed in diazinon and malathion loads. Factors potentially contributing to differences in event loads from individual sites within runoff regimes include time since insecticide application, application rates, number of irrigations/storms since application, differences in irrigation practices, initial placement, and formulation. When expressed on a mass-per-unit-area basis (yield), median chlorpyrifos and diazinon yields in agricultural storm runoff were lower than those reported for the San Joaquin Valley, California (13, 14). Median yields of both compounds in irrigation runoff were higher than those reported for the same area (16). Amounts of organophosphorus insecticides applied to fields and time since application were expected to strongly influence runoff loads. We acquired information on organophosphorus insecticide application to agricultural sites from the California Department of Pesticide Registration Pesticide Use Report (PUR) database (25). Applications of targeted compounds were reported for three agricultural sites both during (A1, A3, and O1) and prior to (A1, A3, and A7) the sampling period (Table 1 and Table S3 in the Supporting VOL. 40, NO. 7, 2006 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2123

FIGURE 3. Summary of dissolved chlorpyrifos, diazinon, and malathion (a) event loads (M) and (b) event mean concentrations (EMCs) for runoff regimes having a sufficient number of events to allow statistical comparisons (Table S6 in the Supporting Information). Abbreviations: ai, agricultural irrigation runoff; as, agricultural storm runoff; rs, residential storm runoff; TEEchl, chlorpyrifos total exposure equivalency (includes contributions from chl, dzn, mal, dichlorvos, dimethoate, and disulfoton; see text). These data are tabulated in Table S4 in the Supporting Information. Information). Applications were not reported for the remaining sites during the sampling period, and we were unable to identify applications to these sites prior to the sampling period. Runoff samples frequently contained organophosphorus insecticides that were not reported to have been applied during or 7-16 months prior to the sampling period. Assuming the PUR database is complete, these additional compounds presumably derived from soil residues of even earlier applications or spray drift from other areas. Although statistical analysis of the relationship between insecticide load and applications or spray drift from other application areas was not possible in most cases, larger loads were generally temporally proximal to applications. For example, during 26 days preceding the April 17, 2000 runoff event, 10.9 kg of diazinon were applied to site A1 (Table S3). This event contributed more diazinon to the stream system than all other measured events combined (9.6 g or 0.09% of these applications). Chlorpyrifos loads in irrigation runoff from site A3 were positively correlated with the amount applied during the preceding 45 days (r2 ) 0.98; p < 0.05). Loads in sampled runoff events represented 0.0002-0.06% of reported applications. The loads observed in this study (as percent of application) were generally smaller than those reported for stream sites in the San Joaquin River Basin, California (0.01-0.27%) (13, 14) and in the Sacramento River, California (0.05-1.3%) (12, 15, 26). Organophosphorus insecticide loads were not consistently correlated with runoff event hydrologic characteristics (see the Supporting Information). The contribution of suspended sediment-associated organophosphorus insecticides appeared important in few sampled events (see the Supporting Information). Event Mean Concentrations. The EMC is a flow-weighted average concentration defined as the ratio of the event 2124

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 40, NO. 7, 2006

pollutant load to the total event runoff volume (27). Use of EMCs facilitates comparison of inputs for sites differing in contributing area. Chlorpyrifos EMCs ranged from less than the detection limit (DL) to 680 ng‚L-1 for agricultural irrigation runoff events and from