concentrations of pollutants in agricultural runoff

VOL.7,NO. 1 WATER RESOURCES BULLETIN FEBRUARY 197 1

CONCENTRATIONS OF POLLUTANTS IN AGRICULTURAL RUNOFF’

Dan M. Wells, Ellis W, Huddleston, and Robert G. Rekers’

ABSTRACT. Eighteen rural lakes in Lubbock County were sampled on a routine basis following runoff-producing rainfall for a period of approximately eighteen months to determine whether or not runoff from intensively farmed agricultural areas contained significant concentrations of nitrates, phosphates, herbicides, or insecticides. An additional fifteen lakes lying within a triangle bounded by the cities of Plainview, Canyon, and Hereford, Texas, were sampled one time during the summer of 1969 to provide additional data regarding the nature and extent of the potential problem in an area with a different soil type and a slightly different cropping pattern.

Based on results of detailed analyses of approximately two hundred samples of water collected from the lakes and an equal number of sediment samples collected from the same lakes at the same time, it appears that the concentrations of all chemical pollutants in runoff from agricultural lands in the High Plains are well below the allowable concentrations for drinking water. (KEY WORDS: *playa lakes; *herbicides; *insecticides; *nitrates; *phosphates; *agricultural runoff)

INTRODUCTION

The High Plains area of West Texas is one of the world’s greatest outdoor laboratories for any type of studies related to agriculture. The area encompasses almost twenty thousand square miles of flat, fertile land, with large areas having similar soil characteristics. The entire area slopes generally from the northwest to the southeast at about ten feet per mile. It is broken by only a very few widely scattered stream beds that drain a very small fraction of the total area. Practically the entire area is farmed intensively under highly mechanized farming practices. The entire area is underlain by the Ogallala formation, which provides irrigation water for approximately six million acres of land.

Because of the absence of a normal drainage network in the area, practically all of the runoff resulting from natural precipitation and from irrigation return flow collects in the playa lakes that occur with a density of about one per square mile throughout the area. Each of these lakes is located in a closed watershed, and it captures all the runoff from the surrounding drainage area, Each lake therefore provides a complete sample of all runoff that occurs from its watershed, and since most of the lakes are dry most of the time, each lake normally provides a self-coniposited sample of each runoff-producing precipitation event.

The total quantity of water collected in the lakes each year is not known with a high degree of accuracy. I t has been estimated by various researchers to be on the order of one million to three million acre-feet.

Paper No. 71012 of the Water Resources Bulletin (Journal of the American Water Resources Associa- tion). Discussions are open until six months from date of publication.

Respectively, Professor of Civil Engineering, Director, Water Resources Center; Associate Professor of Park Administration, Horticulture, and Entomology; Associate Professor of Chemistry; Texas Tech Univer- sity, Lubbock, Texas 79409.

124

CONCENTRATIONS O F POLLUTANTS IN AGRICULTURAL RUNOFF 125

Until recently, playa lakes were considered to be detrimental to the area because of the land they effectively removed from cultivation. Recently, however, as the water table in the Ogallala has declined, area farmers have been more and more interested in putting the playa lake water to beneficial use, either directly on the land or by recharging it into the Ogallala aquifer. Also, because of the declining water table in the Ogallala, playa lakes have come to be regarded as beneficial rather than detrimental attributes of a piece of land.

Because of the historic lack of interest in playa lake water, very little research has been done in the past regarding i t s quality. Since its only significant use has been for irrigation pur- poses, and since experience has shown that it is of adequate quality for that purpose, incen- tives for studying its quality have been lacking. Interest in recharge, along with an increasing awareness of the fact that the lakes provided an ideal full scale outdoor laboratory for determining the significance of agricultural runoff to water quality in general led to the research on which this paper is based.

Scope

Lakes selected for the study were chosen to represent the widest possible variation of test conditions in the area studied. A few of the lakes collect runoff from pasture land only, but most of them receive runoff from cultivated land. Cotton, grain sorghum, soybeans, and wheat are the primary crops grown in the area, and some of these crops were grown in the watersheds of most of the lakes selected.

Some of the lakes studied catch water only rarely and retain it for a very brief period of time following filling. Others receive runoff from fairly small amounts of precipitation and retain the water for much longer periods of time. A few of the larger lakes normally contain water on a year around basis.

EXPERIMENTAL PROCEDURES

Lakes selected for intensive study in the project were plotted on a county map. Insofar as possible the drainage area contributing to each lake was traced out and the total acreage contained in the watershed was estimated. The way in which all land in the watershed was utilized was then determined and plotted on the watershed map. That is, the number of acres of land planted to grain sorghum, cotton, soybeans, wheat, and other crops, and the amount left fallow were determined for each watershed. It is important to note that the total acreage in the watersheds and the total acreage planted to each crop were estimates, rather than precisely calculated acreages. The watershed boundaries are generally so poorly defined that it is impos- sible to determine precisely where they are without detailed field surveys.

An attempt was made to determine the precise quantities of fertilizers and pesticides used in each watershed. This approach proved to be infeasible because of the failure of farmers to keep accurate records of quantities used and because many of the farmers simply could not be located in the time available. The quantities of pesticides and fertilizer used were therefore estimated on the basis of general farming practices in the area.

Sampling Methods

Lakes selected for study were generally sampled after each runoff-producing event, and occasionally between precipitation events.

A one gallon sample of water and a one gallon sample of sediment were taken from each lake on each sampling date. One gallon, brown glass jars which had been thoroughly cleaned

126 Wells, Huddleston, and Rekers

in a chromic acid solution and thoroughly rinsed were used for sample containers. These containers were closed with bakelite screw caps lined with sheets of aluminum foil. Samples collected for analysis were retained in a refrigerator at approximately 2" C until analyzed. Water samples were taken by compositing one-half pint samples from each of sixteen locations within a lake. A one-half pint dipper was used to obtain the sub-samples which were composited directly into the sample jar. Sediment samples collected penetrated approximately the top one inch of sediment below the water-sediment interface, and they were tkane using the same dipper technique.

Pesticide and Fertilizer Use on Watershed

Although exact quantities of pesticide and fertilizer used on each watershed could not be obtained, certain generalizations can be given. Almost all cropland is fertilized with both nitrogen and phosphorus in one of several forms. The total amount of each element applied is fairly uniform from watershed to watershed. Average application rates are 80 pounds of nitrogen and 60 pounds of phosphorus per acre on cotton, and 120 pounds of nitrogen and 60 pounds of phosphorus per acre on grain sorghum. On grain sorghum, however, only about 50 percent of the acreage receives phosphorus fertilizer.

Herbicides are used on 75 to 80 percent of the cropland in Lubbock County. Approxi- mately three-fourths of the cotton and soybean acreage is treated with a preplant application of Treflan (Trifluralin). Other herbicides used on cotton not treated with Treflan are Planivan, Karmex, and a Prometryne. Amiben and Teneran are used on soybeans. Atrazine and Propa- zine are the primary herbicides used on the 75 to 80 percent of the acreage of grain sorghum that is treated.

Insecticides are normally used on cotton and grain sorghum only in response to specific insect problems that arise. In 1969, little of the cotton acreage was treated. Some farmers treated the seed with low rates of an organophosphate systemic insecticide, either Di-Syston or Thimet. Grain sorghum, on the other hand, was subjected to severe infestations of aphids. As a result, almost all the grain sorghum in the area was treated at least once. Parathion or Di-Syston were the primary insecticides used, with about 75 percent of the acres being treated with Parathion.

The estimated size of drainage area, cropping pattern, and quantities of pesticides and fertilizers used on three typical lakes studies are shown in Table 1. Acreages treated as shown in this table were based on the assumption that 10 percent of the cotton and SO percent of the grain sorghum were treated with insecticides and that 80 percent of the cotton, grain sorghum, and soybeans were treated with herbicides. It was also assumed that 90 percent of the cotton, 80 percent of the grain sorghum, and 50 percent of the soybeans were fertilized. Quantities of pesticides and fertilizers shown to be used on each watershed were based on assumed application rates of 0.5 pounds of insecticide per acre of cotton treated, 0.25 pounds per acre of grain sorghum treated, 0.5 pounds of herbicide per acre of cotton, grain sorghum, and soybeans treated, 140 pounds of fertilizer per acre of cotton treated, 180 pounds per acre of grain sorghum treated, and 40 pounds per acre of soybeans treated.

Rtr,iojfarrd Lake Levels

The sampling program was started in December, 1968. Most of the lakes selected for sampling were dry a t that time, and samples were obtained from only three of the lakes. A heavy snow i n March produced a considerable amount of runoff, but six of the lakes selected for study remained dry. A one inch rain in April raised the levels in most of the lakes which had

CONCENTRATIONS OF POLLUTANTS IN AGRICULTURAL RUNOFF 127

been sampled previously and made it possible to collect the first sample from one additional lake .

TABLE 1. Watershed Acreage, Cropping Practices, and Pesticide and Fertilizer Use on Three Typical Watersheds Studied

Lake Number 10 11 12

Watershed Area, Acres Cotton, Acres Grain Sorghum, Acres Fallow, Acres Soybeans, Acres Wheat, Acres Acreage Treated, Insecticide Acreage Treated, Herbicide Acreage Treated, Fertilizer Insecticide Used, Pounds Herbicide Used, Pounds Fertilizer Used, Pounds

800 400 400

0 0 0

240 640 680

70 320

108,000

1,440 320 320 80

640 80

192 1,024

864 56

512 99,200

1,600 320 640 320 320

0 352

1,024 960

96 512

138,880

General rains of almost four inches in early May, 1969, produced runoff t o all but one of the lakes. Additional water was caught in all but one lake as a result of a one and half inch rainfall in the middle of June.

Three lakes had gone dry when a sampling run was made early in July, but a two and one- half inch rain on July 21 produced additional runoff to all lakes but one. This lake failed to catch any runoff during the entire eighteen month sampling program because of a very excel- lent water conservation program practiced on the watershed.

Heavy general rains in September produced runoff to all but five lakes, and a moderate rainfall in November produced additional runoff to several lakes.

No runoff-producing precipitation occurred from November until March, 1970, at which time a moderate rainfall produced runoff to ten of the lakes in the program.

Two sampling runs were made in June and two additional runs were made in July, 1970. No more than ten or twelve lakes contained water at the same time in either June or July, and the water contained in some of the lakes at these times was believed to be derived from irrigation tailwater rather than from runoff from precipitation.

ANALYTICAL PROCEDURES

Approximately 200 samples each of water and sediments were analyzed in performing the research on which this report is based. In general, very low concentrations of pesticides were found in sediments, and all water samples were found to be free of measurable concentrations of pesticides. Exhaustive testing confirmed the ability of the instrumentation and procedures used to measure pesticide concentrations of 0.1 pg/l in water and 0.1 mg/l in sediments. Nitrates and phosphates were found to be present in concentrations that were generally lower


than concentrations naturally existing in groundwater in the area.

Nitrates and Rzosphates

Methods outlined in Standard Methods for the Examination of Water and Wastewater, Twelfth Edition, were used in determining concentrations of nutrients. The procedure recom- mended by Standard Methods was modified in accordance with the procedure suggested by Water Resources Research, Volume 3 , pages 417423, in the analysis for phosphates.

Pesticide Analysis

The analytical procedures required for determination of pesticide concentrations generally

1. The extraction of pesticides from the water or soil, both of which may contain or have

2. Cleanup or separation of the pesticide residues from the extract. 3. Identification and quantitative determination of the concentrations of pesticides. Extraction and cleanup procedures were usually the limiting and the most time consuming

factors when there was organic contamination of the samples. Most of the contamination found was thought to be caused by living or decomposing plant material contained in the water and sediments.

After extraction and cleanup as required, all samples were analyzed on a Varian Aerograph Model 600-C gas chromatograph equipped with a Triden electron capture detector and a Leeds and Northrup Speedomax H-1 millivolt recorder.

Detection of a particular pesticide in an unknown sample requires that the response of the instrument to that sample be compared to the response of the instrument to a known standard pesticide. Provided all variables such as gas flow rate, temperature, electronic variables, type of absorbent column, etc., remained constant, a particular compound passes through the chromatograph at a specific time.

Since all variables involved could not be held absolutely constant from day to day, a procedure was adopted to minimize errors resulting from uncontrollable variables. The procedure adopted was based on the fact that, while absolute detention times for different compounds may vary from day to day as test conditions vary, the relative retention time of all compounds will remain constant under any particular set of test conditions. The instrument was therefore calibrated each day with hexane solution containing a standard concentration of Aldrin, and all other peaks observed were recorded in terms of their retention times relative to Aldrin. Thus,

consisted of three different parts:

previously contained plants.

Hence, the relative retention time for Aldrin is 1.00. Substances with longer retention times than Aldrin have RTA’s greater than 1.00, and substances with shorter retention times in the chromatographic column have RTA’s less than 1 .OO. Relative retention times of several pesticides and some unidentified compounds are shown in Table 2 along with standard devia- tions as determined over a period of approximately five months.

CONCENTRATIONS OF POLLIJTANTS IN AGRICULTURAL RUNOFF 129

TABLE 2. Retention Times for the Non-polar Column

Relative Accepted Relative Retention Times Retention Times

Pesticide Retention Times Retention Times Samples 5% Dow 11 on Standards Minutes (To Aldrin) (To Aldrin) Chromosorb W

Lindane 1.3 f 0.2 0.44 f .01 .44 Heptachlor 2.4 f 0.4 0.78 * .01 .a0 Aldrin 3.0 f 0.4 1.00 f .oo 1 .oo 1 .oo

1.07* 1 SO*

2.15*

2.50*

Die 1 d r i n 6.2 f 0.3 2.00 f .03 2 .oo 1.94

Endrin 6.6 f 1 . 1 2.19 f .05 2.18

pp’ DDT 10.0 * 1.8 3.28 f 0.1 1 3.29 3.29

*Unidentified peaks. Experimental values reported were calculated from chromatograms which were run over a 5 month period. Column Conditions: Support: Chromosorb W 60/80 mesh

Coating: 5% Dow I1 Length: 6 ft. Diameter: 1/8” I.D.

Analytical Bocedure for Water lsamples

The determination of pesticide concentrations in water was a fairly straightforward process usually requiring no cleanup procedure. ‘]The extraction procedure used consisted of agitation of 600 ml of water and 50 ml of N-Hexame by magnetic stirring for 20 minutes in a 1,000 ml Erlenmeyer flask at a rate that resulted in formation of N-Hexane droplets in the water, but not in emulsification. The mixture was then poured into a 1,000 ml separatory funnel and allowed to separate into two layers. The lower water layer was discarded. Tests showed that a single extraction by this method removed 99% + of the pesticides present in the water, thus eliminating the need for further extractions of the sample. The organic layer was then passed through a column containing powdered anhydrous sodium sulfate that had been rinsed previously with N-Hexane. The elutant was collected in a 50 ml amber bottle with a foil lined screw cap.

While repeated tests showed that the method employed was highly effective in extracting pesticides from water, only trace quantities of any type of pesticides were found in any water samples. Trace quantities were found only at the beginning of the mosquito season when spraying for insect control was most intensive, and even then, pesticides were normally found in trace concentrations only if the rainfall had been light enough that extensive dilution had not occurred.


Analytical PLocedure for Sediment

The extraction of pesticides from soil samples was much more difficult and time consuming

1. Twenty-five grams of sediment saturated with water was stirred in a 125 ml Erlenmeyer flask for 20 minutes with 50 ml N-Hexane.

2. Fifteen grams of sediment that had been air-dried and ground to a fine powder was extracted by allowing 25 ml of N-Hexane to gravity filter through the sample.

3. Twenty-five grams of sediment that had been air-dried and ground to a fine powder was extracted by magnetic stirring for 20 minutes with 50 ml of N-Hexane in a 125 ml Erlenmeyer flask.

In all cases, the hexane was collected and passed through a column containing powdered anhydrous sodium sulfate. The extract was then ready for analysis on the gas chromatograph if it was free of interfering organic contaminants.

If the chromatogram obtained indicated the sample contained organic interferences, it was first run through a Florasil cleanup process. In this process, the sample extract was concentrated to approximately 10 ml by evaporation over a 70" C water bath equipped with an aspirated air stream to pull off vapors and speed up evaporation. A 15 gram charge of acti- vated Florasil was tapped in place in a 5/16 inch I.D. column and topped by one inch of anhydrous sodium sulfate. After cooling, the column was pre-eluted with 30 ml N-Hexane and the pre-elutant was discarded. The sample extract was transferred to the column just before the top layer of anhydrous sodium sulfate was exposed to air. The extract was followed with 50 ml N-Hexane and the total volume of elutant was collected and evaporated over a 70" C water bath to a volume of approximately 5 ml. The concentrated extract was then diluted back to pre-Florasil treatment volume and injected into the gas chromatograph.

In extraction method 1, the presence of water tended to produce emulsions which pre- sented a barrier to the passage of pesticides from the soil to the hexane. The emulsions did not break up upon standing nor could they be destroyed by centrifuging. Recoveries by this method were therefore very low and the method was rejected as unacceptable. Of the last two methods mentioned, the former gave slightly better recoveries of pesticides. However, since the second method required more time for gravity filtration and precise collection of the first 25 ml of N-Hexane passing through the filter, the third method was used for most of the work. Recoveries by the method used were comparable to those obtained by the best method, and recoveries in the range of 84% rt 7% to 35% f 4% were obtained, depending upon the type of pesticide and the texture of the soil.

Removal of the organic contaminants extracted from sediments with.out, at the same time, removing pesticides from the extract was the most difficult part of the research. The organic contaminants resulting from decaying plant materials often produced chromatograms that completely masked any pesticides present.

The Florasil cleanup procedure previously described did not always remove all of the trashy contaminants, and it also removed some of the pesticides present. A cleanup method that removed the trashy complex without also removing pesticides was finally found. This method is as follows:

than was the extraction of water samples. Three methods of extraction were tried as follows:

1 . Pipette a 10 ml aliquot of trashy hexane into a 125 ml separatory funnel. 2 . Add 10 ml saturated KOH solution of absolute ethanol to the separatory funnel; shake

3. Leach the ethanol-KOH out of the hexane with 20 ml distilled water, shaking for about for two minutes.

one minute. Allow phases to separate.

CONCENTRATIONS OF POLLUTANTS IN AGRICULTURAL RUNOFF 131

4.

5.

Remove and discard water phase from separatory funnel. Repeat procedure in Step 3 above three times, or until hexane layer is optically clear. Add a small quantity of powdered anhydrous sodium sulfate to remove any water in contact with hexane. The hexane solution is now ready for analysis.

This procedure made possible the quantitative detection of Aldrin with a retention time that fell within the typical range of the trashy complex. The procedure was found to remove Lindane and p, p-DDT and to reduce the concentrations of Treflan. However, since it did not affect Aldrin, Heptachlor, Dieldrin, or Endrin, it was used for the analysis of these pesticides. Other trashy complexes appeared occasionally at higher retention times, and certain water samples also contained these same interfering substances.

RESULTS AND INTERPRETATIONS

None of the water samples analyzed contained measurable concentrations of any of the herbicides or insecticides commonly used in the area. Aldrin, Dieldrin, and DDT were the only insecticides found in sediment samples in the lakes, and no measurable concentrations of herbicides were found in any sediment samples. Measurable concentrations of Dieldrin were found in the sediments in about 80 percent of the lakes. Aldrin was found to be present in sediments in less than 10 percent of the lakes, and DDT was present in detectable concentrations in only three of the samples analyzed.

The average concentrations of nitrates in playa lake waters was found to be 4.1 mg/l . The highest average concentration found in any alke from which more than one sample was taken for analysis was 7.2 mg/l. This concentration is approximately one-sixth the allowable c o n centration for drinking water, and is less than the concentrations found in much of the groundwater in the area.

Concentrations of phosphates found to exist in the playa lakes generally ranged from about 0.1 to about 1 .O mg/l, with an occasional sample containing somewhat more than 1 mg/l.

Results obtained from the analyses of water and sediment obtained from Lake 10 are shown in Table 3. These results are typical of the results obtained from all lakes throughout

TABLE 3. Concentrations of Fertilizers and Pesticides in Water and Sediment Samples Obtained from Lake No. 10

Concentrations in Water Concentrations in Sediments Date NO3 PO, Aldrin Dieldrin DDT

Collected PPM PPM PPM PPM PPM

12-23-68 3-1 8-69 5- 6-69 6-16-69 7-1 6-69 7-2 2 -6 9 9-19-69

1 1-28-69 4-23-70 5-14-70 6- 2-70

22.3 4.3 8.3 4.7 2.2 0.2 0.4 0.3 0.7 3.2 2.9

0.40 0.38 0.90 0.75 0.33 0.33 0.23 0.1 2 0.1 0.6 0.7

N.D. N.D. N.D. N.D. N.D. N.D. N.D. N.D. N.D. N.D. N.D.

0.16 0.1 1 0.2 I 0.052 0.082 0.058 0.12 0.030 0.056 0.032 0.074

N.D. N.D. N.D. N.D. N.D. N.D. N.D. N.D. N.D. N.D. N.D.


the study. Again, it should be noted that none of the water samples contained measurable concentrations of any of the pesticides

CONCLUSIONS

Based on the results of this research, it was concluded that runoff from agricultural lands in the High Plains of West Texas is not a significant source of water pollution.

ACKNOWLEDGEMENTS

The research on which this paper is based was supported by the Federal Water Quality Administration, The Texas Water Quality Board, and Texas Tech University. Most of the sample collection and analytical work was performed by Joe R. Felty, Robert C. Schwartz, Jr. and Janice C. Richards, graduate students in the Department of Chemistry, and Ronald D. Kirby, a graduate student in the Department of Park Admini- stration, Horticulture, and Entomology.

concentrations of pollutants in agricultural runoff

Documents