The Quality of Our Nation's Waters
Pesticides in the Nation's Streams and Ground Water, 1992-2001

---

BACK TO previous page

How the table was made:

This table shows the three most frequently detected unique pesticide mixtures in stream water (percentage of time) or ground water (percentage of samples) from agricultural or urban areas versus the number of pesticide compounds in the mixtures. Only the 25 most commonly detected pesticide compounds in water (see figure 4-2) were considered when identifying unique mixtures.

A unique mixture is defined in this report as a combination of two or more pesticide compounds, regardless of the presence of other compounds that may occur in the same sample (Squillace and others, 2002, p. 1924-1925). For example, a sample containing compounds A, B, and C contains four unique mixtures—AB, AC, BC, and ABC.

The “total detection frequency” threshold (detections at any concentration--some as low as 0.001 micrograms per liter) was used for this table. Six of the 25 most commonly detected pesticides (bentazon, bromacil, carbaryl, 2,4-D, diuron, and norflurazon) cannot be reliably detected when present at concentrations less than 0.1 micrograms per liter. The term “cannot be reliably detected” refers to the inability of the analytical method to detect low concentrations of pesticides when present in water samples. All reported detections of pesticides by NAWQA, regardless of concentration, are “reliable” in that they are conclusively identified by retention time and spectral characteristics. The expected effect of the unreliable detection of these six pesticide compounds when present at concentrations less than 0.1 micrograms per liter for table 5-1 is underestimation of the contribution of bentazon, bromacil, carbaryl, 2,4-D, diuron, and norflurazon to unique mixtures.

Two criteria related to missing values were used in the analysis of mixtures for table 5-1: (1) water samples must have data for at least 80 percent (16 of 20) of the most commonly detected compounds analyzed by gas chromatography/mass spectrometry (GCMS) and must have data for at least 80 percent (4 of 5) of the most commonly detected compounds analyzed by high performance liquid chromatography (HPLC), (2) missing data for individual pesticide compounds in samples that met the criteria provided in (1) were considered nondetections for the purposes of this figure. Analytical methods for pesticides in water are identified in Appendix 1A.

Approximately one third of the stream-water and ground-water sites did not have pesticides analyzed by HPLC. As a consequence, the mixture analysis is based on a smaller number of sites and samples than some of the other topics presented in this report. The expected effect of the missing pesticides described in (2) above for table 5-1 is indeterminate. Only the most frequently detected pesticides are identified in the table. Missing values for frequently detected pesticides would cause underestimation of the frequency of detection of unique mixtures that contain that pesticide and, consequently, may alter the order of the three most frequently detected unique mixtures.

Samples were pooled by sampling medium and land use for analysis. Multiple samples per site (over a 1-year period) were collected to characterize pesticides in stream water, whereas a single sample per well was used to characterize pesticides in ground water.

• For ground water, and for each of the two national land-use classes shown in the table, calculate the number of unique mixtures in each sample:

  Count the number of pesticides detected (d) in each sample. The number of mixtures composed of n pesticides is:

  d! / n! (d – n)!

  where n ranges from 2 to d (Walpole and Myers, 1978, p. 13). The total number of unique mixtures in a sample is the sum of the numbers of unique mixtures calculated for each n from 2 through d.

  A computer program was written to identify the specific pesticide compounds that compose each unique mixture in the sample. The frequency of detection for each unique mixture was determined by dividing the number of samples that contain the unique mixture by the total number of samples in the land-use class. For each land-use class, the three most frequently detected unique mixtures were selected for each n value (the number of pesticide compounds in the mixture) from two through five. In some cases the four most frequently detected unique mixtures were selected to account for ties for third place.

• Stream-water samples were collected more frequently during the pesticide-use season than at other times of the year. Consequently, time-weighted estimates are provided in table 5-1 for stream-water samples to eliminate the bias caused by more frequent sample collection during the pesticide-use season. Time weighting stream-water samples yields an estimate of the percentage of time a unique mixture was detected.

Because of the vast number of unique mixtures, a time-weighting approach based on site-months (rather than sample weights) was used. Unique mixtures in individual stream-water samples were identified using similar procedures as those described above for ground-water samples. The unique mixtures for each sample were pooled by site and month and only the first occurrence of a unique mixture in a site-month was retained for analysis. The frequency of detection of each unique mixture was calculated for each of the national land-use classes, by dividing the number of site-months that contain the unique mixture by the total number of site-months in the land-use class. Mixtures for months without stream-water samples were estimated by assigning the mixtures from the sample nearest in time to the month with missing samples. For each land use class, the three most frequently detected unique mixtures were selected for each n value (the number of pesticide compounds in the mixture) from two through five. In some cases the four most frequently detected unique mixtures were selected to account for ties for third place.

The table was prepared by listing the three (or four, if ties) most frequently detected unique mixtures for mixtures for n values from two through five, by media and land-use class. Listing only the three most frequently detected mixtures would result in a table with empty cells for some unique mixtures that were not detected frequently in another media or land use. Therefore, the empty cells in the table were filled with the calculated frequency of detection for the appropriate mixture, media, and land use. Data in the table are sorted by the number of pesticide compounds in the mixture, the frequency of detection in urban streams, and the frequency of detection in agricultural streams. Frequency of detection was rounded to the one’s place for presentation in the table.

References cited:

Squillace, P.J., Scott, J.C., Moran, M.J., Nolan, B.T., and Kolpin, D.W., 2002, VOCs, pesticides, nitrate, and their mixtures in groundwater used for drinking water in the United States: Environmental Science & Technology, v. 36, no. 9, p. 1923-1930.

Walpole, R.E., and Myers, R.H., 1978, Probability and statistics for engineers and scientists (2d ed.): New York, MacMillan, 580 p.

Pesticide names and analytical methods are presented in Appendix 1.

Information on sampling sites and their characteristics is presented in Appendix 5.

Downloadable concentration data are presented in Appendix 6.

Additional information on analytical approach is presented in Appendix 8.

For more information, contact:

Calculation of unique mixtures:
Jonathon C. Scott
U.S. Geological Survey
202 NW 66 th Street , Building 7
Oklahoma City , OK 73116
voice: (405) 810-4415
fax: (405) 843-7712
email: jon@usgs.gov

Stream water:
Jeffrey D. Martin
U.S. Geological Survey
NAWQA Pesticide Synthesis Project
5957 Lakeside Boulevard
Indianapolis , IN 46278-1996
voice: (317) 290-3333 x148
fax: (317) 290-3313
email: jdmartin@usgs.gov

Ground water:
Paul E. Stackelberg
U.S. Geological Survey
NAWQA Pesticide Synthesis Project
425 Jordan Road
Troy , NY 12180
voice: (518) 285-5652
fax: (518) 285-5601
email: pestack@usgs.gov