An Information Aid for Assessing Possible NRCS Involvement in the State Management Plan Process for Regulation of Pesticides

by Robert L. Kellogg, Joe Bagdon, and Susan Wallace

Natural Resources Conservation Service

with contributions from

Don W. Goss and Joaquin Sanabria

Texas Agricultural Experiment Station, Temple, Texas

February 23, 1998

Resource Assessment and Strategic Planning Working Paper 98-2

Summary

State Management Plans (SMPs) are being proposed by the Environmental Protection Agency (EPA) as a mechanism for regulating five widely used pesticides-alachlor, atrazine, cyanazine, metolachlor, and simazine. Implementation of SMPs will generate a new demand for watershed planning and technical assistance to farmers. The Natural Resources Conservation Service (NRCS) has a role to play by providing farmers with technical assistance in targeted areas, and providing water quality agencies with some of the resource information needed to meet water quality goals.

The purpose of this analysis is to provide insight on which watersheds have the greatest likelihood of needing technical assistance related to SMP implementation. Using the National Resources Inventory, a national-level simulation was conducted of the potential for pesticide loss from farm fields for each of the five pesticides. Maps show which watersheds have the greatest potential pesticide loss from farm fields to exceed water quality thresholds. Nationally, atrazine could potentially exceed drinking water standards at the bottom of the root zone on about 22 million acres (about one-third of the acres treated). The potential is much lower for the other four pesticides-about 600,000-700,000 acres for alachlor and cyanazine, 400,000 acres for simazine, and less than 100,000 acres for metolachlor. One or more of the five pesticides could potentially exceed drinking water standards at the bottom of the root zone on about 23 million acres. Results are also presented for pesticide loss in runoff from farm fields, although the SMP process applies only to ground water. The potential for runoff loss (dissolved fraction) was much greater than for leaching loss in most watersheds. One or more of these five pesticides could potentially exceed drinking water standards in runoff at the edge of the field on about 73 million acres.

The analysis does not show which watersheds are likely to have pesticide contaminated ground water or surface water. Other factors need to be taken into account to assess the potential for water contamination, such as the depth to the water table, characteristics of the vadose zone, microbial activity, and the amount of ground water originating from recharge areas where no pesticides are applied.

Table of Contents

Summary

Background

The Simulation Model

Simulation Results Are Influenced by Pesticide Use Assumptions

How Are Simulation Results Related to Water Quality?

About the Maps and Tables

Assessment of Possible NRCS Involvement in SMPs for Atrazine

Assessment of Possible NRCS Involvement in SMPs for Metolachlor

Assessment of Possible NRCS Involvement in SMPs for Alachlor

Assessment of Possible NRCS Involvement in SMPs for Simazine

Assessment of Possible NRCS Involvement in SMPs for Cyanazine

Using NAPRA as a Decision Aid for Implementing Technical Assistance to Meet Water Quality Goals

Appendix A: Description of the National Pesticide Loss Database

Appendix B: Tables of Acres Where Concentrations Exceed Water Quality Thresholds by NRCS Region and by State

Appendix C: Databases of Acres Where Concentrations Exceed Water Quality Thresholds by Watershed

Note: Each of our maps is also available as a zipped () postscript file and ()PDF file. These files can be used to produce high quality copies of our maps. For more infomation on postscript see our postscript help page. For PDF files Adobe Acrobat is required.

Background

The Environmental Protection Agency (EPA) is proposing to provide state governments with the flexibility to regulate pesticide sale and use to protect ground water quality where warranted by local conditions.¹ EPA's State Management Plan process provides guidelines for states to assume regulatory responsibilities for five pesticides that have been identified as either "probable" or "possible" human carcinogens-alachlor, atrazine, cyanazine, metolachlor, and simazine. The purpose of the program is to prevent ground water contamination that would cause unreasonable risks to human health and to the environment, and to avoid unnecessary regulation in other areas. All five pesticides are broad-spectrum herbicides with extensive agricultural uses. These pesticides have been detected frequently in ground water in many states, sometimes at levels exceeding EPA drinking water standards.

Once the new regulations go into effect, these five pesticides can be used only in accordance with the provisions and requirements of EPA approved State Management Plans (SMPs). The proposed SMP process involves: 1) the identification of sensitive areas that could be vulnerable to pesticide leaching to ground water; 2) monitoring to identify when action thresholds are exceeded; and 3) the implementation of preventative actions to mitigate the contamination. There are several preventative actions that can be taken by states, including education of pesticide users, voluntary adoption of best management practices, mandatory adoption of best management practices, and prohibition of use.

Preventative actions under SMPs will vary depending on where the pesticides are used and the presence of soil and climate characteristics that affect pesticide loss from farm fields. Since the SMP process supports differential management as a solution, it is complimentary to the conservation planning process employed by NRCS. The NRCS conservation planning process includes targeted pest management planning in areas where pesticides have significant potential to move away from the site of application and contaminate water resources. Research has shown that, with careful management, most of the pesticide residues can be prevented from moving into groundwater and surface water most of the time. Research has also shown that some areas of the country and some areas of a watershed contribute more to water pollution than other areas.

NRCS has already been actively involved in the SMP process in some states, providing natural resource data and technical assistance.² These efforts have been successful when affected stakeholders are involved early in the process. NRCS participation in the SMP process provides an opportunity to advance NRCS's conservation mission and help make implementation of SMP's more reasonable and cost-effective for affected producers.

The Simulation Model

Pesticide loss from farm fields was simulated using the National Resources Inventory (NRI) ³ as a modeling framework and as a source of land use data and soil data. The relative environmental risk for each watershed was characterized on the basis of how much the pesticide concentration leaving the field (bottom of root zone or edge of field) exceeded water quality thresholds and the number of acres treated in the watershed. Watersheds used in the analysis are the 2,109 8-digit hydrologic units in the 48 states (917,000 acres average size).

Each NRI sample point was treated as a "representative field" in the simulation model. The simulation was conducted using 13 crops-barley, corn, cotton, oats, peanuts, potatoes, rice, sorghum, soybeans, sugar beets, sunflowers, tobacco, and wheat-which comprise about 170,000 NRI sample points. The statistical weights associated with the NRI sample points are used as a measure of how many acres each "representative field" represents. Land use for the most recent inventory-1992-was used.

Because only 13 crops were included in the analysis (the NRI does not specifically identify any other crops), some areas of the country were not well represented (such as Florida and parts of California). Figure 1 [GIF | Postscript | PDF ] shows the coverage represented by the 13 crops. The yellow areas shown in the map denote areas where less than 40 percent of the cultivated cropland is represented in the simulation model. Most of the cultivated cropland not represented in the simulation model was summer fallow (24 million acres), cropland not planted in 1992 including USDA set-aside acres (13 million acres), other close grown crops (5 million acres), other vegetables (3 million acres), other row crops (3 million acres), and hayland (3 million acres). Non-cultivated cropland (such as orchards and vineyards) and pastureland were also excluded.

The pesticide loss database was created by Don Goss, Texas Agricultural Experiment Station, Temple, Texas, using the GLEAMS process model for 120 soils, 55 climate stations, and 243 pesticides for 20 years of simulated weather. Separate estimates were made for irrigated and non-irrigated cropland. For each set of variables, the concentration of the pesticide at the bottom of the root zone and at the edge of the field was estimated. The concentration was calculated as the total mass of pesticide loss per year divided by the associated water volume per year, and so represents an "annual" concentration. Concentrations were normalized by dividing by the application rate used in the GLEAMS simulations so that results could be applied to actual data on application rates. The 95th percentile concentration obtained over the 20-year GLEAMS simulations was used in this study to represent a worst-case scenario for pesticide loss from farm fields.⁴ (See Appendix A for a description of how the national pesticide loss database was constructed.)

Pesticide use data were taken from Gianessi and Anderson.⁵ Gianessi and Anderson estimated the average application rate and the percentage of acres treated by state for over 200 pesticides and for 84 crops for the time period 1990-93. They organized data from publicly available reports and surveys from Federal and state government agencies. For states and crops not completely covered by the available reports and surveys, Gianessi and Anderson conducted a survey of Extension Service specialists for pesticide use profile information. For states and crops for which there are no published surveys or reports and no expert opinion from specialists, imputations were made by assuming that the State's pesticide use profile was the same as that of a neighboring state. The data base was improved by cross-checking estimates of total pesticide use with other sources of information.

Pesticide use data and pesticide loss estimates were imputed to NRI sample points, and the annual concentrations were compared to water quality thresholds to derive a measure of environmental risk at each NRI sample point. State-level estimates of percent acres treated and application rate were imputed onto NRI sample points by state and crop. Pesticide loss concentrations were imputed onto NRI sample points according to soil type, geographic location, and pesticide. Water quality thresholds chosen for this analysis are EPA drinking water standards: EPA's Maximum Contaminant Level for atrazine (3 ppb), alachlor (2 ppb), and simazine (4 ppb) and EPA's Health Advisory for metolachlor (70 ppb) and cyanazine (1 ppb). The concentration-threshold ratio was calculated for each pesticide at each sample point. If the concentration did not exceed the threshold, risk was set equal to zero.

Watershed-level risk was estimated by calculating Threshold Exceedence Units (TEUs) per watershed. First, acres represented by each NRI sample point were multiplied by the percent acres treated. Then, acres treated were multiplied by the concentration-threshold ratio for each pesticide at each NRI sample point, and summed over all the sample points in the watershed. Thus, TEUs account for the amount the threshold was exceeded and the amount of acres treated in the watershed. TEUs were derived in this manner to compare risk among watersheds. (TEUs are similar in concept to the acre-feet volumetric measure, since they are a multiple of acres times a measure of magnitude at a point.)

Simulation Results Are Influenced by Pesticide Use Assumptions

Estimates of environmental risk are dependent upon assumptions about pesticide use. This analysis was based on state-level estimates of application rates because the database constructed by Gianessi and Anderson is the only consistent, comprehensive pesticide use database publicly available for all regions of the country. However, survey data show that application rates can vary considerably within a state. For example, Gianessi and Anderson estimated that 83 percent of the corn acres in Illinois were treated with atrazine at the average rate of 1.2 pounds per acre. Within Illinois, however, average rates by watershed (8-digit Hydrologic Units) can vary from 0.9 to 1.4 pounds per acre, based on the 1991-1992 Cropping Practice Survey data available from the National Agricultural Statistics Service. For Iowa, average rates of atrazine use by watershed ranged from 0.3 to 1.2 pounds per acre, compared to a state-wide average of 0.9 pounds per acre. For Indiana, average rates of atrazine use by watershed ranged from 1.1-1.7 pounds per acre, compared to a state-wide average of 1.3 pounds per acre.

Figure 2 [GIF | Postscript | PDF ] shows for Iowa, Illinois, and Indiana how the watershed risk scores based on the more specific watershed-level pesticide use estimates from the Cropping Practice Survey database compare to risk scores based on state-level estimates from Gianessi and Anderson. Threshold Exceedence Units per watershed are contrasted for the two pesticide use sources in figure 2 [GIF | Postscript | PDF ]. Most watersheds had lower risk scores when calculated using watershed-level estimates, but some had higher scores. The difference was generally small for most watersheds. Since the Cropping Practice Survey data are available for only a few crops in major producing states, watershed-level estimates of pesticide use could not be used in the national simulation.⁶

How Are Simulation Results Related to Water Quality?

This analysis was designed to show where NRCS needs to be prepared to assist farmers and state water quality agencies with State Management Plans. The analysis does not, however, show which watersheds are likely to have contaminated ground water or contaminated surface water. Other factors need to be taken into account to assess the potential for contamination of ground water, such as the depth to the water table, characteristics of the vadose zone, microbial activity, and the amount of ground water originating from recharge areas where no pesticides are applied. For example, aquifers in some of the areas in the highest category shown in the maps could be protected by impervious layers between the root zone and the aquifer. In other areas, concentrations in water originating from farm fields could be diluted by uncontaminated water originating from noncropland recharge areas. Similarly, surface water contamination is strongly influenced by degradation of pesticides during transport to receiving waterbodies, and the quantity and quality of water originating from noncropland areas.

The following caveats apply to conclusions drawn from this study:

• Analysis does not predict which watersheds will exceed water quality standards. The simulation is only for pesticide loss at the edge of the field and the bottom-of-the-root zone.
• Analysis does not estimate absolute risk. Risk estimates are based on annual concentrations simulated with generalized inputs for pesticide use. Water bodies are subjected to exposures from specific storm events, which can be higher or lower in concentration than the annual concentration.
• Management factors are not included in the simulation. Risk is overstated for watersheds where a majority of farmers are managing the soil to increase organic matter, bulk density, and water holding capacity of the soil, or managing fields in other ways to reduce pesticide loss.
• Risk is based on the 95th percentile concentration simulated over a 20-year weather record, and so represents a "worst case" condition with respect to annual rainfall. Risk estimates per watershed therefore represent potential risk.
• Maps provide a relative ranking of risk among watersheds. It is not known how the class breaks shown in the maps relate to "action thresholds" that trigger SMP preventative actions.
• Only the major crops are included in the simulation. Fruits, nuts and most vegetables are not included.

About the Maps and Tables

Maps were constructed to show which watersheds had the greatest potential for the concentration of pesticide loss from farm fields to exceed water quality thresholds at the bottom of the root zone and at the edge of the field. The maps show areas of the country that are more likely to require technical assistance than other areas. The assessment is based on potential concentrations of pesticides in water leaving farm fields and the number of acres in the watershed where the pesticide has historically been used. Three maps are presented for each of the five pesticides: pounds of pesticide applied, potential for the pesticide to leach below the root zone at concentrations above drinking water standards, and potential for the pesticide to run off the edge of the field (dissolve fraction) at concentrations above drinking water standards.

To facilitate comparisons among the maps, the classes shown in all maps are based on a consistent set of class breaks. Watersheds with the highest ranks are where NRCS needs to be prepared to assist farmers and State water quality agencies with SMP implementation. Farmers in watersheds in the highest category-colored red on the maps-are more likely than farmers in other watersheds to need to adopt alternative management strategies to meet requirements of SMPs. Watersheds in the second highest category have less likelihood of needing technical assistance than watersheds in the highest category, but more likelihood than watersheds in lower categories, and so on. Blue areas of the map have the least potential for contamination on a watershed basis. White areas denote where the pesticide is not used or where the risk is zero.

The class breaks were chosen arbitrarily, but the blue areas are considered to have negligible risk in most cases. Watersheds colored blue have less than 100,000 TEUs. This is equivalent to a pesticide loss concentration being equal to the threshold for 100,000 acres in the watershed. For an average size watershed, 100,000 acres is slightly more than 10 percent of the land base. Although TEU scores less than 100,000 could be associated with a contamination problem in a small subset of the watershed, it is not likely to be associated with concentrations above the water quality threshold for the watershed as a whole.

Because the TEUs per watershed are a function of the number of acres treated, larger watersheds will tend to score higher than smaller watersheds. Watersheds (8-digit HUs) range in size from 16,500 acres to 5,825,300 acres, and average 916, 825 acres, based on the National Resources Inventory. Five percent of the watersheds are less than 241,900 acres, and 5 percent are greater than 1,960,000 acres. For the smallest watersheds, it is possible that scores less than 100,000 TEUs may be associated with undesirable levels of contamination. Additional analysis has shown, however, that this potential for size bias is very minor because the presence or absence of pesticide use is a dominating factor in the calculation.

Estimates of acres associated with threshold exceedences were calculated for states and NRCS administrative regions. Appendix B contains tables with acreage estimates associated with exceedences 1, 2, 3, 5, 7, 10, 15, 20, and 25 times the thresholds. Acreage estimates associated with a threshold exceedence of a multiple of 1 (representing a concentration equal to or greater than the threshold) correspond to the areas shown in the maps. The last two tables in Appendix B contain estimates of acres where one or more of the five pesticides exceeded the MCL or HA.

Assessment of Possible NRCS Involvement in SMPs for Atrazine

Atrazine Use. Of the 13 crops included in the NRI simulation model, Gianessi and Anderson reported atrazine use for corn and sorghum. Based on the Gianessi and Anderson estimates and the crop distribution in the NRI, about 68 million acres of these two crops were treated with atrazine in the early 1990's. The distribution of acres treated by watershed is shown in figure 3 [GIF | Postscript | PDF ]. Atrazine use is highest in Illinois, Indiana, Iowa, eastern Nebraska, and western Ohio, exceeding 150,000 pounds per watershed in many watersheds in these states. Other crops for which Gianessi and Anderson report atrazine use, which are not included in the simulation, are: fallow land, sugarcane, sweet corn, pasture, sod, millet, and seed crops.

Atrazine Leaching Potential. Figure 4 [GIF | Postscript | PDF ] shows the spatial distribution of the potential for atrazine concentrations in leachate to exceed the water quality threshold. Most of the watersheds that have the greatest likelihood of needing technical assistance in support of the SMP process, relative to other watersheds, are in Nebraska, Kansas, Michigan, Wisconsin, Iowa, Illinois, Indiana, Delaware, North Carolina, Georgia, and regions in western Texas and southeastern Missouri. The colored areas of the map represent about 22 million acres, nearly one-third of the acres treated (table 1). About 1.3 million acres could potentially have concentrations in leachate ten or more times greater than the water quality threshold.

Atrazine Runoff Potential. Figure 5 [GIF | Postscript | PDF ] shows the spatial distribution of the potential for atrazine concentrations in runoff at the edge of the field to exceed the water quality threshold. Threshold Exceedence Units for runoff are much higher than for leaching in all regions except portions of the Southeast. Runoff TEUs were over 100 times greater than leaching TEUs in some watersheds (e.g., Iowa and New York). Of the 68 million acres treated with atrazine, 63 million acres could potentially exceed the atrazine water quality threshold at the edge of the field (table 1). About 38 million acres could potentially exceed 10 times the water quality threshold.

Assessment of Possible NRCS Involvement in SMPs for Metolachlor

Metolachlor Use. Of the 13 crops included in the NRI simulation model, Gianessi and Anderson reported metolachlor use for corn, cotton, peanuts, potatoes, soybeans, and sorghum. Based on the Gianessi and Anderson estimates and the crop distribution in the NRI, about 35 million acres of these six crops were treated with metolachlor in the early 1990's. The distribution of acres treated by watershed is shown in figure 6 [GIF | Postscript | PDF ]. Metolachlor use is highest in Iowa, Illinois, and Indiana, exceeding 150,000 pounds per watershed in most watersheds in these states. Other crops for which Gianessi and Anderson report metolachlor use, which are not included in the simulation, are: vegetables, safflower, and sod.

Metolachlor Leaching Potential. Figure 7 [GIF | Postscript | PDF ] shows that there is little potential for metolachlor concentrations in leachate to exceed the water quality threshold in any area of the country. The colored areas of the map represent only about 91,000 acres (table 2), and no watershed had more than 50,000 TEUs. This is in sharp contrast to the relative risk estimated for atrazine, which had a similar use pattern.

Metolachlor Runoff Potential. Figure 8 [GIF | Postscript | PDF ] shows the spatial distribution of the potential for metolachlor concentrations in runoff at the edge of the field to exceed the water quality threshold. As shown for atrazine, Threshold Exceedence Units for runoff are much higher than for leaching, but are still much lower than runoff TEUs for atrazine. Of the 35 million acres treated with metolachlor, 13 million acres (about one-third) could potentially exceed the metolachlor water quality threshold at the edge of the field (table 2). Most of these acres are in Iowa.

Assessment of Possible NRCS Involvement in SMPs for Alachlor

Alachlor Use. Of the 13 crops included in the NRI simulation model, Gianessi and Anderson reported alachlor use for corn, cotton, peanuts, sunflowers, soybeans, and sorghum. Based on the Gianessi and Anderson estimates and the crop distribution in the NRI, about 28 million acres of these six crops were treated with alachlor in the early 1990's. The distribution of acres treated by watershed is shown in figure 9 [GIF | Postscript | PDF ]. Alachlor use is highest in Illinois, Indiana, and western Ohio, exceeding 150,000 pounds per watershed in some watersheds in these states. Other crops for which Gianessi and Anderson report alachlor use, which are not included in the simulation, are dry deans and sweet corn.

Alachlor Leaching Potential. Figure 10 [GIF | Postscript | PDF ] shows that there is little potential for alachlor concentrations in leachate to exceed the water quality threshold in any area of the country. The colored areas of the map represent only about 574,000 acres (table 3), and few watersheds had more than 50,000 TEUs. Like metolachlor, this is in sharp contrast to the relative risk estimated for atrazine leaching.

Alachlor Runoff Potential. In contrast, figure 11 [GIF | Postscript | PDF ] shows that TEUs for alachlor runoff are as high or higher than TEUs for atrazine runoff in areas of the country where alachlor is used. Of the 28 million acres treated with alachlor, 26 million acres could potentially exceed the alachlor water quality threshold at the edge of the field (table 3). About 20 million acres (80 percent of acres treated) could potentially exceed 10 times the water quality threshold.

Assessment of Possible NRCS Involvement in SMPs for Simazine

Simazine Use. Of the 13 crops included in the NRI simulation model, Gianessi and Anderson reported simazine use for corn. Based on the Gianessi and Anderson estimates and the crop distribution in the NRI, about 2 million acres of corn were treated with simazine in the early 1990's. The distribution of acres treated by watershed is shown in figure 12 [GIF | Postscript | PDF ]. Simazine use is highest in Kentucky, Ohio, and Maryland. Other crops for which Gianessi and Anderson report simazine use, which are not included in the simulation, are: alfalfa, fruit, nuts, vegetables, seed crops, and sod.

Simazine Leaching Potential. Figure 13 [GIF | Postscript | PDF ] shows that there is little potential for simazine concentrations in leachate to exceed the water quality threshold in any area of the country. The colored areas of the map represent only 389,000 acres (table 4), and no watershed had more than 50,000 TEUs.

Simazine Runoff Potential. Figure 14 [GIF | Postscript | PDF ] shows the spatial distribution of the potential for simazine concentrations in runoff at the edge of the field to exceed the water quality threshold. Threshold Exceedence Units for runoff are much higher than for leaching, but exceed 100,000 TEUs per watershed in only a few areas (Illinois, Kentucky, Tennessee, Ohio, and New York). Of the 1.7 million acres treated with simazine, 1.6 million acres could potentially exceed the simazine water quality threshold at the edge of the field (table 4). About 400,000 acres could potentially exceed 10 times the water quality threshold.

Assessment of Possible NRCS Involvement in SMPs for Cyanazine

Cyanazine Use. Of the 13 crops included in the NRI simulation model, Gianessi and Anderson reported cyanazine use for corn, cotton, and sorghum. Based on the Gianessi and Anderson estimates and the crop distribution in the NRI, about 21 million acres of these three crops were treated with cyanazine in the early 1990's. The distribution of acres treated by watershed is shown in figure 15 [GIF | Postscript | PDF ]. Cyanazine use is highest in Iowa and Illinois. Sweet corn is the only other crop for which Gianessi and Anderson report cyanazine use, which was not included in the simulation.

Cyanazine Leaching Potential. Figure 16 [GIF | Postscript | PDF ] shows the spatial distribution of the potential for cyanazine concentrations in leachate to exceed the water quality threshold. Most watersheds where cyanazine is used have TEUs below 50,000. Watersheds that have the greatest likelihood of needing technical assistance in support of the SMP process, relative to other watersheds, are in California and Georgia. The colored areas of the map represent about 700,000 acres (table 5).

Cyanazine Runoff Potential. Figure 17 [GIF | Postscript | PDF ] shows the spatial distribution of the potential for cyanazine concentrations in runoff at the edge of the field to exceed the water quality threshold. As shown for the other pesticides, TEUs for runoff are much higher than for leaching in all regions. Runoff TEUs were generally over 1,000,000 per watershed in areas where cyanazine is used the most. Of the 21 million acres treated with cyanazine, 20 million acres could potentially exceed the cyanazine water quality threshold at the edge of the field (table 5). About 19 million acres could potentially exceed 10 times the water quality threshold, and 18 million acres could potentially exceed 20 times the water quality threshold.

Using NAPRA as a Decision Aid for Implementing Technical Assistance to Meet Water Quality Goals

Once a watershed has been targeted for preventative actions under SMPs, more detailed field level models such as NAPRA (National Agricultural Pesticide Risk Analysis) can be used to determine where the most critical areas are within the watershed, and what kinds of alternative management strategies would be required to meet water quality goals. The NAPRA process has been specifically designed to approximate environmental risks of alternative management strategies. NAPRA results estimate relative environmental risks with pesticide toxicity exceedence probabilities and loadings over many years of simulation. NAPRA includes the impacts of climate, water management, soil management, crop management, pesticide management and pesticide toxicity to non-target species. The limits of the evaluation are bottom of root zone and edge of field.

Water quality goals for a watershed will be determined by state and local governments. To achieve these water quality goals, the relatively simple field-by-field approach currently being implemented by NRCS must be expanded to assess the effects of alternative practices on a watershed basis. NAPRA can be used with the Natural Resources Inventory (NRI) or other land use and soils data to identify areas within a watershed that need special emphasis pest management planning. Field office staff in these areas can then consult with producers and crop consultants to ascertain existing practices, and to promulgate economically feasible alternatives. NAPRA can be used again to test the alternatives to determine if they are adequate to meet water quality goals for the watershed. Using NAPRA to target technical assistance to high risk areas within high risk watersheds is the only practical way to substantially improve water quality with limited resources.

NAPRA implementation has two phases: 1) development of NAPRA Pesticide Loss Databases by national and state level specialists, and 2) NAPRA reports generated from these data that quantify the relative environmental risks associated with each conservation alternative. NAPRA implementation requires a team effort. NRCS seldom has all the expertise necessary to fully parameterize the model, so it is imperative to partner with other agencies, private industry, the agricultural community, and the environmental community.

The ultimate goal is to help farmers choose pesticide management alternatives that reduce hazardous pesticide losses in environmentally sensitive areas. Traditionally, agricultural producers have made pesticide management decisions based on efficacy and economics. Under SMPs, it will be necessary to factor environmental risk into the decision making process. Adoption of practices such as reductions in pesticide use, improved efficacy of pesticide applications through integrated pest management, matching pesticide and management practice selection to site conditions, and the use of pesticides that are less toxic to the environment will be necessary. Using models to estimate the environmental benefits of these practices can help agricultural producers make more informed decisions that meet the environmental goals of conservation planning.

Appendix A: Description of the National Pesticide Loss Database

Appendix B: Tables of Acres Where Concentrations Exceed Water Quality Thresholds by NRCS Region and by State

Appendix C: Databases of Acres Where Concentrations Exceed Water Quality Thresholds by Watershed