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

Temporal Trends of Selected Agricultural Chemicals in Iowa's Groundwater, 1982-95: Are Things Getting Better?

Dana W. Kolpin, Debra Sneck-Fahrer, George R. Hallberg, and Robert D. Libra


RESULTS

The overall frequency of detection for the agricultural chemicals under investigation for all groundwater samples collected for the IGWM during the months of April to September (1982-95) is given in Table 2. Most reported concentrations were found to be below current U.S. Environmental Protection Agency's maximum contaminant levels or health advisory levels for drinking water, with nitrate by far having the most common exceedences (Table 2).

An examination of all groundwater samples collected for the IGWM during the months of April to September from 1982-95, identified a significant (P = 0.009, Kruskal-Wallis test) temporal trend in frequency of detection only for metolachlor (Table 3). The frequency of detection for metolachlor progressively increased between the 1982-86 and 1992-95 time periods.

To determine if temporal patterns in the detection of agricultural chemicals were masked by confounding factors, the data also were grouped by well depth and aquifer type. For well depth, the wells were divided into three groups; shallow (</= 15 m), intermediate (16-30 m) and deep (>30 m) wells. Well depth has been used in previous studies as a rough surrogate for groundwater age (e.g., Kalkhoff et al., 1992; Kross et al., 1990). The hypothesis relating to well depth is that temporal changes in detection frequency may be most evident in more recently recharged (younger) water. Grouping by well depth revealed significant temporal trends in the frequency of detection for the following: shallow wells (atrazine; P = 0.027, Kruskal-Wallis test), intermediate wells (metolachlor; P = 0.006, Kruskal-Wallis test), and deep wells (no significant temporal trends) (Table 4).

To evaluate the significance of aquifer type with respect to the detection of agricultural chemicals, the wells were divided into three groups representing alluvial, glacial-drift, and bedrock aquifers. Because of their differing geology, groundwater age likely differs significantly between these aquifers. The hypothesis relating to aquifer type is that temporal changes in detection frequency may be most evident for aquifer types having the youngest water. Significant temporal trends in the frequency of detection were identified for the following: alluvial aquifers (metolachlor; P = 0.023, Kruskal-Wallis test), glacial-drift (no significant temporal trends), and bedrock aquifers (cyanazine; P = 0.023, Kruskal-Wallis test) (Table 5).

To ensure that the sampling design (for example, changing population of wells sampled for a given year) of the IGWM has not caused misidentifications of temporal trends in the data, a subset of 89 municipal wells (Fig. 2) that were sampled repeatedly from 1982-1995 was selected for a more in-depth investigation of temporal patterns. The advantage of this subset of wells is that temporal changes in median chemical concentrations can be examined rather than just frequencies of detection because the same population of wells are examined through time. An examination of this subset of groundwater samples identified a significant temporal trend in median concentration only for atrazine, with a significant decrease (P = 0.021, Kruskal-Wallis test) being determined. This temporal pattern was caused by a substantial decrease in median atrazine concentration between the 1987-91 and 1992-95 time periods. There was no significant change (P = 0.905, Kruskal-Wallis test) in median atrazine concentration between the 1982-86 and 1987-91 time periods. About 80% (27 of 34) of the wells where atrazine was detected had decreases in median atrazine concentration between the 1987-91 and 1992-95 time periods.

Because of the availability of paired observations for this subset of 89 wells, the Wilcoxon signed-rank test also may be used to test for temporal changes in median concentrations. For this statistical test, the median 1992-95 concentrations were subtracted from the median 1987-91 concentrations for each well. If no data were available for the 1987-91 time period for a particular well, the median 1992-95 concentration was subtracted from the median 1982-86 concentration. Examining these differences in paired observations identified a significant decrease in median atrazine concentrations (P <0.001, one-tailed Wilcoxon signed-rank test) and a significant increase in median metolachlor concentrations (P = 0.042, one-tailed Wilcoxon signed-rank test) (Fig. 3). Those agricultural chemicals where no significant difference was identified had roughly equal positive and negative concentration differentials (Fig. 3).

The decreasing trend in median atrazine concentrations is consistent with results of other long-term research in groundwater and for streams in the midwestern United States. Concentrations of agricultural chemicals have been measured in groundwater from the 267 km² Big Spring basin (Libra et al., 1992; Rowden, et al., 1995; Iowa Department of Natural Resources, Geological Survey Bureau, 1996, unpublished data) on a weekly basis since 1982. The flow-weighted mean atrazine concentrations from the Big Spring basin were 0.40 µg/L (1982-86), 0.65 µg/L (1987-91), and 0.20 µg/L (1992-95). Bedrock aquifers in northeast Iowa (Floyd and Mitchell Counties) were sampled from a network of 25 wells in May 1986 and again in May-June 1994 (Quade et al., 1994). The frequency of atrazine detection decreased from 72% to 64% in these wells between the two sampling periods. The mean atrazine concentration, however, showed an even more dramatic trend, decreasing from 1.8 µg/L in 1986 to 0.25 µg/L in 1994. A study of two areas in central and southeastern Wisconsin also found an apparent temporal relation between decreasing atrazine use and decreasing atrazine concentrations in groundwater (Saad, 1997). A study of 50 midwestern streams in the United States sampled during May-June for the years 1989, 1990, 1994, and 1995 found a statistically significant decrease in atrazine concentration between the 1989-90 and the 1994-95 sampling periods (Donald A. Goolsby, U.S. Geological Survey, written commun., 1996). Decreased concentrations in streams are important to aquifers having strong groundwater/surface water interactions (e.g. alluvial aquifers). Research has shown that streams can be a source of atrazine contamination to alluvial aquifers (Squillace et al., 1993; Wang and Squillace, 1994; Squillace, 1996).

As when the entire IGWM data set was examined, temporal patterns in concentrations of agricultural chemicals may be masked by confounding factors, such as well depth and aquifer type. Grouping by well depth revealed significant temporal trends in the median concentration for the following: shallow wells, atrazine -- significant decrease (P = 0.041, Kruskal-Wallis test; P<0.001, one-tailed Wilcoxon signed-rank test), metolachlor -- significant increase (P = 0.022, Kruskal-Wallis test; P = 0.031, one-tailed Wilcoxon signed-rank test); intermediate wells, atrazine -- significant decrease (P = 0.049, one-tailed Wilcoxon signed-rank test); and deep wells (no significant temporal trends). The temporal variability for atrazine and metolachlor decreased with increasing well depth (Fig. 4). Decreased chemical variability with increasing well depth is consistent with the literature (Hallberg, 1987; Hallberg, 1989b; Kolpin and Thurman, 1995). As expected, changes in chemical applications or climatic conditions at the land surface should be reflected by changes in chemical concentrations most rapidly in the shallowest (youngest) groundwater.

Grouping by aquifer type revealed significant temporal trends in the median concentration for the following: alluvial aquifers, atrazine -- significant decrease (P = 0.039, Kruskal-Wallis test; P <0.001, one-tail Wilcoxon signed-rank test), metolachlor -- significant increase (P = 0.039, Kruskal-Wallis test); glacial-drift (no significant temporal trends); and bedrock aquifers (no significant temporal trends) (Fig. 5). In other investigations, alluvial aquifers have been determined to be more susceptible, in general, to contamination by agricultural chemicals than the generally deeper bedrock aquifers (Hallberg, 1989a; Kolpin et al., 1994; Kolpin and Goolsby, 1995). The relative absence of overlying low-permeability material, generally local flowpaths with surface-recharge areas in proximity to wells, and the relatively rapid rates of groundwater movement are all likely contributing factors to the temporal patterns found in alluvial aquifers. However, the relatively small number of wells completed in glacial-drift and bedrock aquifers for the subset of IGWM wells examined decreased our ability to adequately determine temporal patterns for these types of aquifers.

Although the temporal changes in median metolachlor concentrations determined for this study appear minor (fewer wells affected) in relation to those of atrazine (figs. 3-5), the results are equally important. The increase in median metolachlor concentrations that occurred in water samples during the 1992-95 time period might be the leading edge of an upward trend of metolachlor contamination in Iowa's groundwater. Continued collection of water-chemistry data for the IGWM is required to determine if this represents a long-term increase in metolachlor concentrations in groundwater.

Precipitation and Chemical Use

Two major factors that can have a temporal effect on chemical concentrations in groundwater are variations in precipitation (recharge) and chemical use patterns (Hallberg and Keeney, 1993; Hallberg et al., 1993; Lucey and Goolsby, 1993; Barbash and Resek, 1996). There is no apparent relation between precipitation and the temporal patterns in median concentrations found for atrazine and metolachlor. In general, 1982-86 was a time of relatively normal precipitation, 1987-91 included the driest consecutive two-year period on record (1988-89) in Iowa (Kross et al., 1990), and 1992-95 included the single wettest year on record (1993) in Iowa (Kolpin and Thurman, 1995; Wahl et al., 1993). An improved understanding of the relation between precipitation and chemical concentrations in groundwater may be obtained if data on rainfall and/or recharge in proximity to sampled wells were available. Water level measurements from sampled wells also would be useful in determining possible changes in recharge for an individual well. However, all the wells in this study are municipal wells. Few municipalities allowed access for water-level measurements to be made or collected data on water levels themselves.

Although there appears to be a slight decrease in the statewide total mass applied and intensity of use for nitrogen fertilizer (Table 1), no significant temporal trends were noted in either the frequency of detection or median nitrate concentrations in groundwater. There are several possible explanations for this: (1) the general decrease in nitrogen-fertilizer applications are not large enough to cause a significant decrease in groundwater concentrations, (2) more time is required for this decrease in applications to be reflected by groundwater concentrations, (3) other sources of nitrate contamination, such as animal manure, are not taken into account, (4) the statewide application data may not adequately characterize actual changes in fertilizer applications around the sampled wells, and (5) nitrate occurrence and concentrations in shallow groundwater are sensitive to changes in precipitation and recharge because of the mobility of nitrate (Hallberg and Keeney, 1993; Hallberg et al., 1993). Other studies in Iowa have suggested that nitrate concentrations increased in the 1992-95 period, in part, because of pre-existing drought conditions that allowed considerable nitrogen to accumulate in the soil even though fertilizer applications had been reduced (Lucey and Goolsby, 1993; Rowden et al., 1995).

No significant temporal trends were noted in either the frequency of detection or median alachlor concentrations in groundwater, even though statewide alachlor use has decreased about 60% (Table 1) from 1982 to 1995. There are possible explanations for the absence of a temporal trend in median alachlor concentrations in groundwater. First, alachlor was found relatively infrequently in the wells having the longest record of water-chemistry data (7 of 89). Thus, because 92% of the wells did not have alachlor concentrations above analytical reporting limits the identification of a significant temporal pattern becomes more difficult. Alachlor has been found to degrade rapidly in the soil zone (Clay et al., 1995; Potter and Carpenter, 1995; Thurman et al., 1996), with a major degradation product being alachlor ethanesulfonic acid (alachlor-ESA; 2-[(2,6-diethylphenyl)(methoxymethyl)amino]-2-oxoethanesulfonic acid). Alachlor-ESA is a much more persistent compound and has been detected in groundwater almost 10 times as frequently as alachlor (Kolpin et al., 1996; Kolpin et al., 1997). If long-term data on alachlor degradation products were available, a more definitive relation between alachlor use and chemical concentrations (parent + degradation products) in groundwater might have been identified. Second, alachlor has recently shown a dramatic decrease in statewide use. The statewide use of alachlor has decreased from first among all herbicides in 1985; to fourth in 1994; to twelve in 1995 (George R. Hallberg, University of Iowa Hygienic Laboratory, written commun., 1996). This sharp decrease in alachlor use that has occurred in recent years may take additional time to be reflected in measured concentrations from the few wells that were found to have alachlor in groundwater.

The temporal pattern of decreasing median atrazine concentrations determined for the subset of 89 wells examined for this study appear coincident with the changes in statewide atrazine use (Table 1) and are suggestive of a causal relation. The estimated mass of atrazine being used in Iowa has remained relatively constant from 1982-1995, with only a slight decrease (about 12%) in use occurring (Table 1). However, there has been about a 40% decrease in the intensity of use (Table 1) that also could affect the amount of chemical transport to groundwater. Research has shown that application rate can be an important factor in determining the amount of chemical transport (Cheng, 1986; Ng et. al., 1995; Rhode, 1981). The decrease in the intensity of atrazine use can be attributed to several factors: 1) the increased use product mixes/tank mixes containing atrazine that reduce the mass of atrazine applied (per unit area) by using it in combination with other herbicides, 2) the increase in postemergent applications that use lower rates of atrazine application (Hartzler and Wintersteen, 1991), and 3) State restrictions of the rate, per hectare, at which atrazine can be applied within selected areas in northeast and north-central Iowa (Iowa Department of Natural Resources, 1994). The relatively slow degradation rate for atrazine (Nair and Schnoor, 1992; Kruger et al., 1993; Clay et al., 1995; Widmer and Spalding, 1995) allows more parent compound to be transported to groundwater, and thus, may cause more apparent relations between chemical use and chemical concentrations in groundwater (more data above analytical reporting limits available for the temporal analysis).

No significant temporal trends were noted in median cyanazine concentrations in groundwater, even though cyanazine use has decreased about 25% (Table 1) from 1982 to 1995. Similar to alachlor, cyanazine has been found to degrade rapidly in the soil zone (Muir and Baker, 1976), and is found infrequently in Iowa's groundwater (Detroy et al., 1988; Kross et al., 1990). A major degradation product of cyanazine is cyanazine amide [2-chloro-4-(1-carbamoyl-1-methylethylamino)-6-ethylamino-s-triazine]. Cyanazine amide is a much more persistent compound and has been detected in groundwater about two to three times more frequently than cyanazine (Kolpin et al., 1996; Kolpin et al., 1997). If long-term data on cyanazine degradation products were available, a more significant relation between cyanazine use and chemical concentrations (parent + degradation products) in groundwater might have been identified.

The temporal pattern of increasing frequency of metolachlor detections and increasing median metolachlor concentrations identified through this study appear coincident with the changes in statewide metolachlor use (Table 1) and are suggestive of a causal relation. The mass of metolachlor used has increased over 50% from 1982 to 1995 (Table 1). The combination of the large mass of metolachlor used in Iowa (Table 1) and its relative mobility (Bowman, 1990; Wietersen et al., 1993; Kruger et al., 1996) may have contributed to the temporal trends identified for metolachlor in this study. The agricultural-chemical data summarized for this study are the basis for continued assessment of the identified temporal trends to evaluate their significance. An improved understanding between chemical use and chemical concentrations in groundwater may be obtained if data on actual chemical use in proximity to sampled wells were available.


Next: Discussion

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/jeq26/res.html
Page Contact Information: gs-w_nawqa_whq@usgs.gov
Page Last Modified: Tuesday, 04-Mar-2014 14:44:29 EST