National Water-Quality Assessment (NAWQA) Project

 Go to:      NAWQA Home

Pesticide National Synthesis Project

Home Publications National Statistics Data Pesticide Use Water-Quality Benchmarks PNSP Internal

Pesticides in Streams of the United States--Initial Results from the National Water-Quality Assessment Program

By Steven J. Larson, Robert J. Gilliom, and Paul D. Capel

U.S. GEOLOGICAL SURVEY
Water-Resources Investigations Report 98-4222
Sacramento, California, 1999


PESTICIDE LOAD IN RELATION TO USE

The transport of pesticides in different types of streams was discussed previously in terms of yield, or load per unit area. A wide range in annual pesticide yields was observed, with estimates ranging more than three orders of magnitude for the agricultural sites (fig. 19). Much of this variability is caused by differences in the amounts of pesticides used in the agricultural basins. Comparisons of loads among the basins can be made more effectively by expressing the annual pesticide load as a percentage of the annual use in the drainage basin.

The general relation between pesticide load and pesticide use is shown in figure 29. These plots show the estimated annual stream load of each pesticide in relation to the amount of that pesticide used agriculturally in the basin. All target compounds with reported agricultural use of 100 kilograms (kg) or more in a basin are included. For basins with low pesticide use, represented by points on the left side of the plots in figure 29, the percentages were highly variable, ranging more than three orders of magnitude. Many of these points represent small agricultural basins for which accurate load estimates are difficult to obtain, as discussed previously. Moreover, the pesticide-use estimates for small basins are probably much less reliable than for larger basins. Points on the right side of the plots in figure 29, which show results from integrator sites and the larger agricultural sites, are probably a more accurate representation of the relation between pesticide use and pesticide load in streams.

The estimated load for many compounds, both herbicides and insecticides, was between 0.01 and 1.0 percent of the amount applied in the basin. The points along the bottom of the plots in figure 29 show that at some sites there was no measurable load (all nondetections were assumed to be zero) for a number of compounds despite relatively high use in the basin. The large variability in the values for load as a percentage of use is to be expected for this diverse group of compounds which have considerable variability in physical properties and in application practices (Larson and others, 1995). In addition, the sites represent basins with a wide range of soil types, climate, and topography, all of which influence the runoff of pesticides to streams.

Figure 29 shows that insecticide loads generally represent a somewhat smaller percentage of use than herbicide loads. This is shown more clearly in figure 30, in which the total load of the target herbicides and insecticides is plotted in relation to the total amount of the target herbicides and insecticides used agriculturally in the basins. Data used for figure 30 are tabulated in table 7. Similar to figure 29, there is considerable scatter in the data in figure 30 for basins with low use. The total load of the target herbicides in basins with high herbicide use, however, generally represents between 0.1 and 1 percent of the total amount of the target herbicides used in the basins. The total load of the target insecticides is lower than the total herbicide load at most sites, representing 0.01 to 0.1 percent of the amount of the target insecticides used in the basins. The median values of the percentages calculated for each site are 0.52 for total herbicides and 0.04 for total insecticides (table 7). For comparison, data from several much larger basins are included in figure 30. These data show the relation between load and use in the Mississippi River Basin and several large subbasins from a 1991 study (Larson and others, 1995). The general agreement between the results from the 1991 study and the results from the NAWQA sites indicates that the relation between load and use is relatively constant over a wide range of spatial scales.

The percentages of pesticide use shown in figure 30 are based on the relation between the load of total (summed) herbicides and insecticides and their summed agricultural use and thus represent an average value for the target compounds. The percentage of pesticide use for a specific site is most influenced by those pesticides with the greatest use in the basin of that site. For example, the load of total herbicides in the Platte and White rivers was about 1 percent of the total amount of the target herbicides applied in their drainage basins. The Platte and White River basins are primarily in corn-growing areas, and the herbicides atrazine, metolachlor, and cyanazine accounted for 70 to 75 percent of total herbicides use. In contrast, the load of total herbicides in the Red River and Fargo (redn-rr.fargo) and at the Emerson (redn-rr.em) was about 0.1 percent of the total amount of the target herbicides applied in the basins of these sites. The main crops grown in both of the Red River basins are wheat and other grains and relatively small amounts of corn. The primary herbicides (of the target compounds included in this report) used in the basins of the Red River sites are trifluralin and EPTC. Atrazine, metolachlor, and cyanazine account for less than 25 percent of total herbicide use in the basins of the two Red River sites. As shown below, the loads of EPTC and trifluralin in streams consistently represent a smaller percentage of the amount of pesticides used in the basin than the loads of atrazine, metolachlor, and cyanazine; thus it is important to consider which specific pesticides are used when comparing loads of total herbicides or insecticides among basins.

The relation between load in streams and agricultural use in the basins of the streams is shown in figure 31 for several pesticides. Data used for figure 31 are tabulated in table 7. Loads of the herbicides atrazine, metolachlor, and cyanazine (fig. 31A-C) represent about 1 percent of the amounts of these compounds used in the drainage basins. Percentages of trifluralin and EPTC (fig. 31D,E) are lower than the percentages for atrazine, metolachlor, and cyanazine by 1 to 2 orders of magnitude, with loads representing about 0.01 to 0.1 percent of the amounts of trifluralin and EPTC used in the drainage basins. This large difference can partly be attributed to differences in the physical properties of these herbicides and their methods of agricultural application. Trifluralin and EPTC are considerably more volatile than atrazine, metolachlor, and cyanazine and generally are incorporated into the soil as they are applied, reducing the potential for transport in surface runoff. Atrazine, cyanazine, and metolachlor are commonly applied to the soil surface before crops have emerged from the soil, increasing the likelihood of transport in surface runoff. In addition, trifluralin has a strong tendency to become attached to soil particles, further reducing its potential for transport in surface runoff. Data from the 1991 study of the Mississippi River and several of its large tributaries (Larson and others, 1995) are included in figure 31A-E. Values of loads as a percentage of use for the large rivers sampled in the 1991 study are consistent with the data from the integrator sites and from most of the agricultural indicator sites sampled during the current study. The general agreement between the results from the 1991 study and the results from the current study indicate that the relation between load in the streams and agricultural use in the basins is relatively constant for these herbicides among basins with widely varying use of these compounds.

Stream load was also compared to basin use for two of the most commonly detected insecticides, carbaryl (fig. 31F) and carbofuran (fig. 31G). Variability was higher for these insecticides than for the herbicides, but the load at sites with high use of carbaryl and carbofuran in the drainage basins was relatively consistent at about 0.1 percent of the amount used in the basins. Determination of an accurate estimate of load for the insecticides is more difficult than for many of the herbicides because of the generally low detection frequencies for insecticides, especially at the agricultural indicator sites and the integrator sites. In addition, values of load as a percentage of use are potentially less accurate for some insecticides because of their widespread use in nonagricultural applications. Nonagricultural use is not included in the estimates used in figure 31.

Several sources of potential error should be noted for the estimates of annual load and for the values of load as a percentage of use:

(1) As mentioned previously, load estimates probably are biased low for small basins because of the low probability of sampling peak concentrations in small streams.

(2) For compounds with low detection frequencies, load estimates may be biased low because concentrations less than the detection limit were set equal to zero for the load calculations. In some cases, pesticides may have been present in the stream at concentrations high enough to affect the load total but less than the concentrations required for detection.

(3) Load estimates were made only for parent compounds and no degradation products were considered.

(4) Estimates of pesticide use were inferred from data on crop acreage and from average application rates of pesticides on specific crops. For small basins in particular, a change in either one of these can have a large effect on the accuracy of the use estimates.

(5) Nonagricultural use is not accounted for in the pesticide-use estimates. The integrator sites were selected to represent a variety of land uses, and it is very likely that significant nonagricultural use of specific pesticides occurs in each of these basins.

(6) The same estimates of pesticide use were used for the calculations of load as a percentage of use regardless of the year of sample collection. If the pesticide load varies significantly from year to year at a site, estimates of load as a percentage of use also will vary. Because the load is influenced by stream discharge, weather, agricultural practices, and other factors (Leonard, 1990), some variation in pesticide load from year to year is to be expected in a stream even if the same amounts of pesticides were applied in the basin. This variability is illustrated in figure 32, in which total herbicide and insecticide loads are compared for 25 sites where load estimates were available for similar periods during different years. The diagonal line in these plots represents 1:1 agreement. The axes in these plots are logarithmic so a substantial deviation from the 1:1 line means that the load at a site was quite different for the 2 years. For total herbicides (fig. 32A), loads were similar for the 2 years at most sites. Total herbicide loads for the 2 years were within a factor of 2 at 14 of the 25 sites and within a factor of 10 at all but 1 site. The relatively constant loads of total herbicides for these 2 years suggest that calculation of load as a percentage of use on the basis of 1 to 2 years of sampling is reasonable for many of the target herbicides, especially since values for a number of different sites are available. For total insecticides (fig. 32B), differences between the estimated loads for the 2 years generally were greater. Total insecticide loads for the 2 years were within a factor of 2 at 4 of the 25 sites and within a factor of 5 at 16 of the 25 sites. At four sites, loads during the 2 years differed by more than a factor of 10. These differences may reflect actual differences in the amounts of insecticides used in the basins during different years, as well as differences owing to the six factors mentioned above. Insecticide use is inherently more variable than herbicide use because many insecticides are used in response to specific pest problems rather than on a preset schedule. Because the same estimates of insecticide use were used for both 1993 and 1994, values of insecticide load as a percentage of use are potentially less accurate than the values calculated for herbicides.

Accessibility FOIA Privacy Policies and Notices

Take Pride in America home page. FirstGov button U.S. Department of the Interior | U.S. Geological Survey
URL: http://water.usgs.gov/nawqa/pnsp/pubs/wrir984222/load.html
Page Contact Information: gs-w_nawqa_whq@usgs.gov
Page Last Modified: Tuesday, 04-Mar-2014 14:44:38 EST