The National Agricultural Pesticide Risk Analysis (NAPRA) provides a tool to assess the relative pesticide risk based on the likelihood of pesticide loss exceeding certain environmental criteria. NAPRA uses the USDA environmental fate model GLEAMS (Groundwater Loading Effects of Agricultural Management Systems) to estimate the probability of pesticide loading and concentrations from a crop field. The NAPRA process considers climate, soils, pesticide properties (i.e. toxicity and persistence), tillage practices and field slope as well as slope length. Given regional climate, the process can be used to identify soils and tillage which are susceptible to pesticide losses. In our NAPRA analysis, annual four-day maximum and annual concentrations of pesticides in runoff and in percolation, probabilities of occurrence of pesticide concentration exceeding EPA’s health advisory level, and a relative risk index were estimated. The acute and chronic risk to surface and groundwater are rated from the NAPRA scenarios. This study is to implement the NAPRA approach in Michigan to assess the relative pesticide risk, which helps identify alternative management to minimize environmental risk. The risk-based approach also helps to prioritize surface and groundwater pesticide monitoring programs. Two Michigan counties, nine soils, two herbicides, and different application rates and methods were used in the NAPRA runs. Results in Clinton county showed that among the predominant agricultural soils, Capac loam had a higher potential risk to surface water contamination when a pesticide such as atrazine was applied, compared to Capac sandy loam and Marlette soils. Marlette sandy loam appeared to have a higher risk to groundwater quality due to higher pesticide leaching in the soil. In Cass county where soils are predominantly sandy, the potential of pesticide leaching to groundwater is high. Surface application increases the pesticide loss in runoff while soil incorporation increases the loss in percolation. Another deliverable of this study is an Internet web site of NAPRA implementation, which has been created to make the information more accessible. The NAPRA web page (http://www.iwr.msu.edu/~ouyangda/napra) can be accessed by farmers, extension personnel and other interested users to assess the potential risk of pesticide loss for a specific soil, pesticide and tillage.

risk assessment, nonpoint source pollution, pesticide, water quality modeling, NAPRA, GLEAMS

The use of pesticides in food and fiber production has brought about environmental concerns, one of which is the potential for contamination of surface and groundwater. It is important to evaluate the potential risk of pesticide application on croplands posed to the environment, particularly in water resources. Pesticide losses from agricultural lands are affected by climate, soil and pesticide properties as well as management practices. Some pesticides may travel far from where they were applied and reach drinking water wells, streams and rivers. In Michigan, herbicides are detected in many drinking water wells located in shallow water aquifers (MSU and WMU 1994). Studies conducted by the U.S. Geological Survey (USGS) showed that following application of herbicides to croplands, there are significant amounts of herbicides flushed into streams by later spring and summer rainfall. This may produce high herbicide concentration for weeks or months in surface waters (Goolsby and Battaglin, 1995).

Water quality issue is an important part in watershed management. Agricultural nonpoint source pollution including pesticides applied in agricultural croplands has gained great attention in recent years. Minimizing pesticide contamination while maintaining high agricultural production is a challenge. Evaluating the potential risk, identifying and managing those affecting factors as well as monitoring water quality are important efforts in reducing pesticide risk. The National Agricultural Pesticide Risk Analysis (NAPRA) provides a tool to assess the potential risk of pesticide application in croplands. NAPRA is one of two official NRCS-USDA methods for determining the complex environmental risks of pesticide use. NAPRA is the second and third tier of NRCS’s three-tiered method for assessing the off-site risk of pesticide movement (Bagdon et al 1994). By comparing different scenarios, existing management practices and alternatives which have minimal environmental risk can be identified, and the target pesticide and sites can be selected for water quality monitoring programs. In this study, NAPRA procedures were used for relative risk assessment for two Michigan counties.

The purpose of this paper is to provide an example of the implementation of NAPRA. This includes climate-based probabilistic analysis of the risk of pesticide losses that may potentially reach surface and groundwater. It gives an insight into identification of site-specific best management practices which may significantly reduce pesticide risk. It also provides a risk-based approach to selecting the priority pesticides and sites for water quality monitoring programs.

This paper focuses on several major agricultural soils in two Michigan counties, Clinton and Cass county (Fig. 1). Clinton county is located in the South Central Michigan.  According to the county soil survey (USDA 1978), Capac and Marlette soils are predominant soils, accounting for approximately 40% of total acreage in the county. The Capac series are nearly level to gently sloping, and somewhat poorly drained soil. The Marlette soils are gently sloping to steep, well drained and moderately drained. Annual precipitation in the county is about 30 inches of which approximately 60% is received during the growing season (May - October). Corn, soybeans and wheat are the major field crops in the county.

Cass county is located in the Southwestern Michigan.  The county soil survey (USDA 1991) indicates that Kalamazoo loam is the predominant soil in the county (30%),  followed by Oshtemo sandy loam (15%) and Spinks loamy sand. The Kalamazoo soils are formed in loamy and sandy material, and well drained. Permeability in these soils is moderate in the upper part of the profile and rapid in the lower part. The Oshtemo soils are also formed in loamy and sandy material with moderately rapid permeability. The Oshtemo series are well drained soils. Spinks soils are formed in sandy material, and well drained. These soils are moderately rapidly permeable or rapidly permeable. Annual precipitation in the county is about 33.6 inches of which approximately 19.7 inches is received during the growing season (May - October).

The National Agricultural Pesticide Risk Analysis (NAPRA) tool developed by the NRCS-USDA was used in the study. The USDA environmental fate model GLEAMS (Groundwater Loading Effects of Agricultural Management Systems, Keisel et al 1992) is the core model in NAPRA to estimate the probability of pesticide loading and concentrations from a crop field. The data required in GLEAMS are also required in NAPRA. The NAPRA procedure takes into account a variety of environmental factors including climate, soil and pesticide properties (i.e. toxicity and persistence), tillage practices and history, field slope and slope length. Among NAPRA results, annual four-day maximum concentrations of pesticides in runoff and in percolation generated by the model are used to assess the acute (or short-term) risk to surface and groundwater, respectively. Annual concentrations of pesticides in runoff and in percolation are used to assess the chronic (or long-term) risk to surface and groundwater, respectively. Relative risks were rated based on the U.S. Environmental Protection Agency’s (EPA) health advisory levels (HAL) or maximum contamination level (MCL).

NAPRA runs require a variety of data including soil, climate, pesticide, crop/tillage management practices and erosion factors. These data are stored in databases for running the NAPRA scenarios.

Soil: Michigan soil data for Clinton and Cass counties were obtained from the NRCS State Soil Survey Database (SSSD) provided by the NRCS Michigan state office. A program called Prepsoil (Hesketh et al 1994) was used to calculate soil attribute values from the SSSD. Soils were used in this study is shown in the following table.

Table 1. Selected soils in Clinton and Cass county, Michigan

Climate: Climate data were downloaded from the Climate Data Access Facility System (CDAFS) maintained by the National Water and Climate Center, USDA Natural Resources Conservation Service. At least 30 years of climate data are recommended to calculate probabilities on a statistical base (Plotkin et al 1994). Forty-six years of climate data from 1949-1995 were used to run NAPRA for Clinton county. Forty-one years of climate data from 1954-1994 were used for Cass county. Data include daily precipitation and temperature, monthly average maximum and minimum temperature. The missing data were replaced by the data from the nearest weather stations.

Crop/tillage management: Questionnaires, interview and telephone inquiries were conducted with County Extension and NRCS staff to obtain crops, tillage, pesticide application in the area. Soil Conservation District staff in the St. Johns field office of Clinton and in Cass county provided most of the information. These data were used as the baselines for crop/tillage management practices in the study areas. Both the baselines and modified crop/tillage data were used to generate the NAPRA scenarios.

Pesticides data: Pesticide properties data was provided NRCS/ARS/CES Pesticide Properties Database (Hornsby 1995) with the NAPRA package. Pesticide properties include toxicity and persistence. A pesticide with a longer half-life may cause long-term water quality problems while a pesticide with high water solubility may pose a greater risk to groundwater, particularly for a shallow aquifer. These properties of pesticides were cited from literature (Hornsby 1995) (Table 2).

Table 2. Properties of selected pesticides

Data for pesticide application including rates, timing, and methods were provided by the NRCS-USDA field office in the study areas.

The EPA’s drinking water standards (lifetime health advisory levels) were used in the risk rating procedure. The probabilities of pesticide concentrations in runoff and percolation which exceed the EPA’s drinking water standard were used to rate the risk level for surface and groundwater. The maximum 4-day and annual concentrations were used to assess the potential acute and chronic risks. Four risk levels (low, medium, high and very high) were used in the risk ratings (Table 3).

Table 3. Risk rating criteria

The risk reflects the potential for surface and groundwater contamination caused by pesticide loss from croplands rather than the actual risk in the water supply. The dilution of pesticide in water is not considered in the current NAPRA model. In addition to four risk level ratings, a relative risk index was calculated from the ratio of the projected pesticide concentration to EPA’s health advisory levels.

Five Clinton county soils and four Cass county soils were input in NAPRA runs. Results for two herbicides (atrazine and metolachlor) are included in this paper. The application rate in active ingredient was 1.12 kg/ha (1 lb./acre) for atrazine and 1.68 kg/acre (1.5 lb./acre) for metolachlor. Two different application methods are soil incorporation and surface application. Probabilities of occurrence that herbicide concentration exceeds EPA’s health advisory level were calculated based on the NAPRA outputs. These outputs include annual 4-day maximum concentration and annual concentration in percolation and runoff. The results are shown in Table 4 and 5.

Table 4. Occurrence probability (%) of atrazine concentrations exceeding EPA’s health advisory level and risk ratings. (Application with soil incorporation and surface application for corn, active ingredient = 1.12 kg/ha)

Note: L – Low; M – Medium; H – High; VH – Very High.

Table 5. Occurrence probability (%) of metolachlor concentrations exceeding EPA’s health advisory level and risk ratings. (Application with soil incorporation and surface application for soybean, active ingredient = 1.68 kg/ha)

Note: L – Low; M – Medium; H – High; VH – Very High.

In addition, a relative risk index was calculated which indicates how many times the projected pesticide concentration in the field is higher than EPA’s health advisory level. As an example, the estimated relative risk index with different application rates for Kalamazoo soil (applied on surface) is shown in figure 2, and two herbicides with the same application rate for Marlette loam soils in figure 3.

Fig.2 Probability of relative risk for atrazine loss in percolation (annual) from Kalamazoo soils

Fig.3 Probability of relative risk for atrazine and metolachlor loss in runoff (annual) from Marlette loam soils (rate = 1.12 kg/ha active ingredient)

In Clinton county, soils in hydrological group C (Capac soils) have a relatively higher potential of pesticide loss in runoff than soils in hydrological group B (Marlette soils). Among the five soils, Capac loam has the highest risk of pesticide loss in runoff. In Cass county, groundwater is more vulnerable to pesticide losses and application of pesticide in those soils have high risk ratings for pesticide loss in percolation. This is due to the high content of sands and low soil organic matter. For surface water quality, although the potential risk for long-term water quality is low, the risk for acute contamination may be very high.

Among manageable practices, application methods can be used to reduce pesticide losses. Application methods may be associated with tillage practices. For example, surface application is used in no-till while soil incorporation may be used with some conservation tillage. In no-till, runoff and sediment yield are reduced. This leads to reducing pesticide load in runoff, but not the concentration in runoff (Pantone et al 1996). Data in Table 4 and 5 show that pesticide concentration in runoff increased when pesticides were applied to the soil surface, compared to soil incorporation which increases pesticide concentration in percolation. This suggests that there may be a trade-off in reducing pesticide loss in runoff and percolation.

Pesticide movement and transport are affected by both the timing and intensity of rainfall. Pesticide concentration in runoff is greater when a rainfall event occurs soon after application (Pantone et al 1996). The application rate also affects pesticide losses. Hanson et al (1997) found that there is a positive linear relationship between atrazine application rate and mass of atrazine in drainage water from a silt loam soil. They reported that the concentration of atrazine in drainage water increased 1/3 as application rate increased 1/8 in that soil. By reducing application rate, probability and frequency of pesticide concentration exceeding EPA’s health advisory levels can be decreased. This can be seen from the curves in figure 2 which show that with application decreases, the relative risk index decreases accordingly, meaning the probability and severity of pesticide contamination in groundwater are decreased.

NAPRA scenarios show that in many soils, groundwater is more vulnerable to pesticide contamination, particularly to atrazine. It suggests that for water quality monitoring programs, presence of atrazine in groundwater may need to be monitored, among others.

NAPRA results included in this paper are part of the NAPRA implementation project. More results can be browsed through our NAPRA web page (http://www.iwr.msu.edu/~ouyangda/napra). The Internet provides an efficient and easy means to access this NAPRA information which is also integrated into a web-based interactive conservation planning project. Delivering research outcomes through today’s information technologies is promising. The Institute of Water Research homepage in which the NAPRA web site resides has more than 600 "hits" each day, although the number of "hits" in the NAPRA web page itself has not been counted.

In general, clay loam and loam soils have a greater chance of pesticide loss in runoff, while sandy loam soils have a greater risk to percolation loss. Reducing application rate can directly reduce pesticide losses. Atrazine has a higher risk than metolachlor in all soils because of its higher toxicity. Surface application of pesticide generally increases the chance of pesticide loss in runoff, which poses a greater risk to surface water, while soil incorporation may increase pesticide loss in percolation. There is a trade-off in managing practices to protect surface and groundwater quality.

Although total load of pesticides is not discussed in this paper, both total load and concentration of pesticide loss may play a role in water quality. Results from other studies showed that reducing one indicator may not necessarily reduce the other (Pantone et al 1996). Therefore, further research is needed in this aspect and the implications for conservation tillage and no-till practice in which pesticides are usually surface-applied at a greater rate.

The risk-based analysis from NAPRA provides a useful approach for prioritizing pesticide monitoring programs by selecting the target pesticides and sites with high risk potentials. For disseminating and implementing the NAPRA approach, there is a great potential by means of the World Wide Web information technology.

We are grateful for the financial support from the National Water and Climate Center (WCC), NRCS-USDA for this research. Special thanks go to the NAPRA Team of the NRCS-USDA at Massachusetts for their technical assistance. Jim Squires and Steven Law, the NRCS staff in Clinton county, and Alex Bozymowski, Jr. in Cass county helped us in collecting crop, tillage and pesticide data. We also thank Jim Marron for allowing us to download the climate data from the WCC. Joe Ervin provided helpful information for pesticide detection in groundwater in Cass county. Tom Moen and Elaine Brown in the Institute of Water Research at Michigan State University reviewed the manuscript.

Bagdon, J., E. S. Hesketh, S. Plotkin, and M. S. Hugo (1994) NAPRA Technology Transfer Overview: An Introduction. NAPRA Technical Notes.

Goolsby, D.A., and W.A. Battaglin (1995) Occurrence and Distribution of Pesticides in Rivers of the Midwestern United States. In: Leng, M.L., E.M.K. Leovey, and P.L. Zubkoff (eds.) (1995) Agrochemical Environmental Fate – State of the Art. Lewis Publishers.

Hanson, J.E., D.E. Stoltenberg, B. Lowery, and L.K. Binning (1997) Influence of Application Rate on Atrazine Fate in a Silt Loam Soil. J. Environ. Qual. Vol.26, No.3. pp#829-835.

Hesketh, E.S., J. Bagdon, and S. Plotkin (1994) Prepsoil and RV Soil Attributes in the NAPRA Process. NAPRA Technical Notes.

Hornsby, A.G., R.D. Wauchope, and A.E. Herner (1995) Pesticide Properties in the Environment. Springer-Verlage New York, Inc.

Keisel, W.G., R.A. Leonard, and F.M. Davis (1992) GLEAMS Users’ Guide.

Michigan State University (MSU), and Western Michigan University (WMU) (1994) Cass County Studies – Executive Summary. Research report.

Pantone, D.J., K.N. Potter, H.A. Torbert, and J.E. Morrison, Jr. (1996) Atrazine Loss in Runoff from No-tillage and Chisel-tillage Systems on a Houston Black Clay Soil. J. Environ. Qual. Vol.25, No.3. pp# 572-577.

Plotkin, S., J. Bagdon, and E.S. Hesketh (1994) Climate and Irrigation in the NAPRA Process. NAPRA Technical Notes.

USDA (1991) Soil Survey of Cass County, Michigan.

USDA (1978) Soil Survey of Clinton County, Michigan.

Go Back