Article pubs.acs.org/est
Assessment of the Effects of Farming and Conservation Programs on Pesticide Deposition in High Plains Wetlands Jason B. Belden,†,* Brittany Rae Hanson,† Scott T. McMurry,† Loren M. Smith,† and David A. Haukos‡ †
Department of Zoology, Oklahoma State University, Stillwater, Oklahoma, United States Department of Natural Resources management, Texas Tech University, Lubbock, Texas, United States
‡
S Supporting Information *
ABSTRACT: We examined pesticide contamination in sediments from depressional playa wetlands embedded in the three dominant land-use types in the western High Plains and Rainwater Basin of the United States including cropland, perennial grassland enrolled in conservation programs (e.g., Conservation Reserve Program [CRP]), and native grassland or reference condition. Two hundred and sixty four playas, selected from the three land-use types, were sampled from Nebraska and Colorado in the north to Texas and New Mexico in the south. Sediments were examined for most of the commonly used agricultural pesticides. Atrazine, acetochlor, metolachlor, and trifluralin were the most commonly detected pesticides in the northern High Plains and Rainwater Basin. Atrazine, metolachlor, trifluralin, and pendimethalin were the most commonly detected pesticides in the southern High Plains. The top 5− 10% of playas contained herbicide concentrations that are high enough to pose a hazard for plants. However, insecticides and fungicides were rarely detected. Pesticide occurrence and concentrations were higher in wetlands surrounded by cropland as compared to native grassland and CRP perennial grasses. The CRP, which is the largest conservation program in the U.S., was protective and had lower pesticide concentrations compared to cropland.
■
INTRODUCTION The United States High Plains grasslands and wetlands have been extensively altered over the past century.1,2 Cultivation agriculture has converted over 15 million ha of grassland to cropland and caused widespread unsustainable sedimentation of the dominant depressional playa wetlands.3−5 This has drastically impacted the natural goods and services provided by these ecosystems to society.6 Despite our knowledge of the impacts of row crop agriculture on ecology of depressional wetlands, few studies have investigated the level of pesticide contamination in these wetlands or the potential impacts of pesticide residues on resident biota. Conservation programs have been implemented in the High Plains, primarily by the U.S. Department of Agriculture (USDA), to reduce negative agricultural impacts. The dominant USDA program in the High Plains is the Conservation Reserve Program (CRP). The original goal of the program was replacement of highly erodible cropland to perennial cover. The highest density of CRP land in the nation is in the High Plains with over 3 million ha planted to perennial grasses, primarily exotics but also some native mid and tall grass planted in historical short-grass prairie zones.6 Indeed, over $97 million is spent annually on CRP payments in the High Plains.7 Additionally, a second USDA program, the Wetland Reserve Program (WRP), is common in some areas such as the Rainwater Basin in central Nebraska and is focused on preserving © 2012 American Chemical Society
and restoring hydrological function of wetlands. Typical treatments include sediment removal and placement of an easement of perennial grasses around the wetland.8 As depressional wetlands, most runoff, including water, associated sediments, and potentially contaminants such as pesticides, are deposited in playa basins. However, the existing residue levels and spatial distribution of chemicals in playas of the High Plains is unknown. As primary sites of biodiversity provisioning and recharge of the nation’s largest aquifer, the deposition of contaminants in playas has important direct and indirect societal impacts. Therefore, we measured pesticide levels in playa sediments embedded in the dominant land-use types in the western High Plains and the Rainwater Basin in south central Nebraska (cropland, CRP or WRP, and native grassland or reference condition). Two hundred and sixty four playas, randomly selected from the three land-use types, were sampled from Nebraska and Colorado in the north to Texas and New Mexico in the south. Sediments were examined for most of the commonly used agricultural pesticides applied throughout the region. Our approach was to analyze sediment samples as they represent a primary sink for much of the pesticide load in Received: July 15, 2011 Accepted: February 22, 2012 Published: February 22, 2012 3424
dx.doi.org/10.1021/es300316q | Environ. Sci. Technol. 2012, 46, 3424−3432
Environmental Science & Technology
Article
Analysis of Samples. Twenty grams of each composite sample was ground to dryness with 40−60 g of anhydrous sodium sulfate and extracted using dichloromethane as a solvent in a Soxhlet apparatus (6 h, greater than 24 rinses). Following extraction, extracts were evaporated to less than 10 mL and solvent exchanged to hexane. The reduced extract was eluted through a commercially prepared florisil column (0.5 g Sigma Aldrich, Atlanta GA) followed by 10 mL of hexane:ethyl ether 1:3. This combined eluate contained the pesticides, leaving many interfering compounds on the column. The volume of the clean extract was reduced to 1 mL of hexane. A second aliquot of each sample was weighed and heated at 105 °C for 24 h to determine the percent solid of each sample. All concentrations are reported on a dry-mass basis. Separation of the analytes by gas chromatography was performed on an Agilent 6850 GC (Agilent, Palo Alto, CA) with a 30 m × 0.25 mm HP-5 column (Agilent) and splitless inlet. The oven was programmed to start at 70 °C, hold for 1.0 min, ramp at 10 °C/min to 160 °C, ramp at 3.5 °C/min to 255 °C, ramp at 3.0 °C/min to 290 °C and hold for 2.0 min. The inlet temperature was 240 °C and the transfer line was 290 °C. The column flow was 1.0 mL/min (37 cm/sec average velocity). Detection and quantitation of analytes were conducted by mass spectrometry on an Agilent 5975c inert source instrument (Agilent, Palo Alto, CA). Electron ionization was used (70 eV) as the ionization source. Temperature of the source was 230 °C and the quadropoles were at 150 °C. Detection was based on 3-ion selected ion monitoring (SIM; SI Table 2). Calibration was conducted using internal standards (decachlorobiphenyl, atrazine D-5, and tributylphospate). Quantitation limits (QL) are based on the lowest calibration standard. Each QL was at least 3× the method detection limits calculated by measuring low-level concentrations of analytes in extracts of clean soil. QLs are reported instead of method detection limits to ensure that low level hits in samples are quantitatively accurate and reduce qualitative uncertainties by ensuring that all quantitative and qualitative ions are measurable (SI Table 2). Quality control was performed throughout sampling and analysis. Field/travel blanks were performed at a 5% frequency of samples as were laboratory blanks. Field blanks were generated by adding reference sediment mixture to a jar in the field, while laboratory blanks were generated by extracting reference sediment at the same time samples were extracted. Laboratory matrix spikes and matrix spike duplicates (using clean reference sediment) were also performed at 5% frequency and at a concentration of 10 μg/kg. Surrogates (triphenylphosphate, trichloromethylxylene, and p-terphenyl) were added to each sample to monitor accuracy of each extraction/analysis (10 μg/kg spike level). Data Analysis. Data were heavily “left-censored” due to a high frequency of values (>50% in most cases) reported at the quantitation limit. Thus, standard descriptive statistics including mean and standard error would be biased.10 Frequency above the quantitation limit (QL) was calculated and comparisons were made among land-use groups using contingency tables and χ2 analysis (the probability of making a Type I error (α) = 0.05). Percentile ranks at 75th, 90th, and 95th percentiles were calculated using rankings and not assuming any standard distribution (percentile function, Excel, Redmond, CA). For example, the 50th percentile is the median. Since the data were heavily censored, nonparametric statistical approaches were used for comparisons.10 Statistical comparisons among land-use
depressional wetlands, many of which are dry for the majority of the year. We report pesticide concentrations found in playas and compare those values to relevant and sensitive toxicological end points. In addition, we examined the influence of surrounding land-use and USDA conservation programs on the concentration of pesticides in playa sediments.
■
MATERIALS AND METHODS Materials. Neat standards (>98%) of all analytes were purchased from Sigma-Aldrich (St. Louis, MO). All solvents and other reagents were pesticide or GC/MS grade (Burdick and Jackson, Muskegon, MI). Site Selection. A total of 264 playas was selected and sampled across five states. Based on pesticide usage patterns, cropping patterns, and climatic differences, playas were split among three regions including southern playas (western Oklahoma, eastern New Mexico, and western Texas), northern playas (western Kansas, western Nebraska, and northeastern Colorado), and the Rainwater Basin (south-central Nebraska). Figures 1−3 in the Supporting Information (SI) illustrate the geographical range of the playas. Playas were classified according to the dominant land-use in their immediate vicinity (500 m zone). For cropland playas, the presence of each crop type within the 500 m buffer zone was determined for each site using appropriate data layers from Cropscape (United States Department of Agriculture; http://nassgeodata.gmu.edu/ CropScape/; accessed December 2011). In the south, 156 playas were sampled and proximal land-use determined (63 cropland, 48 CRP, and 45 native grassland). Cotton was the most common crop occurring at 64% of cropland sites followed by winter wheat (54%), field corn (13%), and grain sorghum (10%). In the north, 66 playas were sampled (23 cropland, 21 CRP, and 22 native grassland). Winter wheat was the most common crop occurring at 83% of cropland sites followed by field corn (35%), grain sorghum (17%), and millet (4%). In the Rainwater Basin 42 playas were sampled (15 cropland agriculture, 15 WRP, and 12 reference wetlands). Reference wetlands are the wetlands remaining in this intensively farmed region that retain most of their natural function as determined through the Hydrogeomorphic Method. These were classified as such by Nebraska Game and Parks Commission personnel. Field corn and soybean were the dominant crops occurring at 93 and 53% of cropland sites, respectively. Sediment Collection. Sediments were collected as a composite sample by sampling from the top 6 cm in three random locations within each wetland (∼500 g total sample). Samples were kept on ice in chemically clean glass jars until arrival at the laboratory where they were frozen until analysis. Most wetlands were dry at the time of sampling. Texas, New Mexico, and Oklahoma samples were collected in June and July of 2008, while Kansas, Colorado, and Nebraska samples were collected in June and July of 2009. Analyte List. Pesticides were chosen as analytes based on use within the region, feasibility for analysis using the available techniques, and potential toxicity. SI Table 1 lists the analytes along with their chemical abstract numbers and selected toxicity end points. Insecticides were emphasized based on their potential toxicity. In addition, several fungicides were added for the northern playas due to increased trends in usage and potential toxicity.9 All of the compounds on this list are extractable with organic solvent and can be directly analyzed using gas chromatography coupled with mass spectrometry allowing a single measurement technique. 3425
dx.doi.org/10.1021/es300316q | Environ. Sci. Technol. 2012, 46, 3424−3432
Environmental Science & Technology
Article
Table 1. Summary Data for Pesticide in Southern Playa Sediments (TX, OK, NM)a number of samples total
above QL
QL1
freq. >QL, %
max. value, μg/ kg
median > QL, μg/ kg
percentiles for all samples, μg/kg2 75
37 37 2.3 25
95
