JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (8)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Selim, H. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Selim, H. M.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Selim, H. M.
Related Collections
Right arrow Surface Water Quality
Right arrow Solute Transport Models
Right arrow Kinetics
Right arrow Agricultural Pesticides
Journal of Environmental Quality 32:1058-1071 (2003)
© 2003 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America

TECHNICAL REPORTS
Surface Water Quality

Retention and Runoff Losses of Atrazine and Metribuzin in Soil

H. M. Selim*

Agronomy Dep., Louisiana State Univ. AgCenter, Baton Rouge, LA 70803

* Corresponding author (mselim{at}agctr.lsu.edu)

Received for publication May 15, 2002.

   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Minimizing herbicide runoff and mobility in the soil and thus potential contamination of water resources is a national concern. Metribuzin [4-amino-6-(1,1-dimethylethyl)-3-(methylthio)-1,2,4-triazin-5(4H)-one] and atrazine [2-chloro-4-ethylamino-6-isopropylamino-1,3,5-triazine] dynamics in surface soils and in runoff waters were studied on six 0.2-ha sugarcane (Saccharum spp.) plots of a Commerce silt loam (finesilty, mixed, superactive, nonacid, thermic Fluvaquentic Endoaquept) during three growing seasons under different best management practices. Metribuzin was applied in the spring as a postemergence herbicide and atrazine was applied following winter harvest. Both herbicides were applied on top of the sugarcane rows as 0.6- or 0.9-m band width application, or broadcast application, where the entire area was treated. Maximum effluent concentrations were measured from the broadcast treatment and ranged from 600 to 1100 µg L-1 for atrazine and 250 to 450 µg L-1 for metribuzin. Atrazine runoff losses were highest for the broadcast treatment (2.8–11% of that applied) and lowest for the 0.6-m band treatment (1.9–7.6%), with a similar trend for metribuzin losses. Measured extractable herbicides from the surface soil exhibited a sharp decrease with time and were well described with a simple first-order decay model. For atrazine, estimates for the decay rate ({lambda}) were higher than for metribuzin. Results based on laboratory adsorption–desorption (kinetic–batch) measurements were consistent with field observations. The distribution coefficients (Kd) for atrazine exhibited stronger retention over time in comparison with metribuzin on the Commerce soil. Moreover, discrepancies between adsorption isotherm and desorption indicated slower release and that hysteresis was more pronounced for atrazine compared with metribuzin.

Abbreviations: MRM, multireaction model


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
CONTAMINATION of surface and ground water from applied agricultural chemicals is a national concern. Specifically, the potential for contamination of water supplies by pesticides, as a consequence of extensive agricultural use, is a major environmental concern. Atrazine and metribuzin are two major herbicides that are used extensively in sugarcane production in Louisiana. Chemical weed control programs for sugarcane usually require two herbicide applications, one in early spring and another before the crop canopy closes. Sugarcane producers in southern Louisiana refer to the latter application as the layby treatment. Herbicides at layby are applied following the last cultivation, usually in late May or June, as broadcast and directed underneath the sugarcane canopy. Shading and competition from sugarcane plants usually control weed development after canopy closure. Recently, metribuzin has been recommended as a postemergence treatment for controlling seedling johnsongrass [Sorghum halepense (L.) Pers.], other seedling grasses, and most broadleaf weeds in sugarcane. Atrazine is recommended as a postemergence treatment to control winter or early spring weeds as well as a preemergence treatment for layby or fallow fields.

Atrazine has been detected in ground water samples collected from agricultural lands. Though atrazine concentration levels in the ground water tend to be low, its long persistence suggests that it reaches the water table, and is for that reason subject to accumulation in ground water (Wehtje et al., 1984; Isensee et al., 1990; Johnson et al., 1995). The USEPA lifetime health advisory level for atrazine in drinking water is 3 parts per billion (µg L-1). Metribuzin is characterized by its high water solubility, moderate half-life (40 d), and low affinity or adsorption capacity to the soil (Koc = 60). Here Koc represents the ratio of the distribution coefficient Kd to the amount of organic carbon (OC) present in the soil (Wauchope et al., 1992). The USEPA lifetime health advisory level for metribuzin in drinking water is 200 µg L-1.

It is estimated that of all the acreage of sugarcane grown in Louisiana, about half receives atrazine and metribuzin as part of the annual production practices (Gianessi and Puffer, 1991). These application practices combined with the high annual rainfall of southern Louisiana could result in significant amounts of herbicide reaching nontarget sites in the environment. In a field study of herbicide fate in southern Louisiana soils, Southwick et al. (1992) reported effluent outflow atrazine concentrations of 82 to 403 µg L-1 in subsurface drains within 5 to 11 d after application on a Sharkey clay soil (very-fine, smectitic, thermic Chromic Epiaquert). In a subsequent study, Southwick et al. (1995) found high concentrations of atrazine (81 µg L-1) and metribuzin (94 µg L-1) within 8 to 10 d after application. In addition, atrazine concentrations in the surface Sharkey soil of 1432, 503, 208, 93, and 49 µg L-1 were found 2, 14, 35, 73, and 84 d after application, respectively. Southwick et al. (1995) and Johnson et al. (1995) suggested that preferential or macropore flow was a significant process in this Sharkey clay soil.

In the fall of 1997, the Louisiana Department of Agriculture and Forestry informed the Louisiana Department of Environmental Quality (LDEQ) about problems with atrazine and triazine pesticides in Bayou Plaquemine, the drinking water source for a portion of Iberville Parish, Louisiana. Levels of atrazine above the USEPA standard of 3 µg L-1 had contaminated the surface drinking water supply for this area. The source of the atrazine has been related to agricultural runoff from sugarcane and corn (Zea mays L.) production in that area. As a result, a cooperative effort between LDEQ, through USEPA (Section 319), and the Louisiana Agricultural Experiment Station was initiated to implement best management practice (BMP) demonstration projects that prevent atrazine from reaching the drinking water intakes or significantly lower the concentration of atrazine in the surface water. Support for this work was through the USEPA Clean Water Act (Section 319) and the Coastal Zone Management Act (Section 6217) with the primary purpose to evaluate, demonstrate, and implement best management strategies to improve water quality in watersheds nationwide.

To reduce amounts of herbicides susceptible to runoff leaching losses, various forms of best management practices are recommended. For example, several best management practices have shown distinct advantages of minimum and no-till systems in reducing soil losses and runoff of applied chemicals (Dao, 1991, 1995; Banks and Robinson, 1982). In this study, the focus was to investigate the effect of reducing herbicide amounts applied to the soil (grams of herbicide applied per acre) without reducing their efficacy. To this end, the effect of broadcast versus band application of atrazine and metribuzin on their runoff amounts and concentrations in the effluent was investigated.

In Louisiana, a sugarcane crop is typically planted in rows 1.8 m apart. It was assumed that band application on top of sugarcane rows, which requires less herbicide on an acre basis, would result in lower amounts of herbicides susceptible to runoff losses. More significantly, runoff amounts of herbicides would be further reduced because of subsequent retention by the (untreated) surface soil in the sugarcane furrows. In this study, the amount of herbicide applied, on an acre basis, ranged from one-half to one-third of that applied for the broadcast treatment. It is recognized that understanding management strategies requires an assessment of the retention characteristics of these herbicides in the soil. Therefore, the first objective of this research was to study the dynamics of atrazine and metribuzin in surface soils and runoff waters in sugarcane fields under different best management practices. A second objective was to study the sorption–desorption of atrazine and metribuzin and characterize their kinetic behavior in soil. Such information is a prerequisite to minimize herbicide runoff and mobility in the soil and thus potential contamination of water resources.


   MATERIALS AND METHODS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Field Investigation
The major goal of this field study was to compare losses of the applied herbicides atrazine and metribuzin and the insecticide azinphosmethyl [dimethyl S-((4-oxo-1,2,3-benzotriazin-3(4H)-yl)methyl) phosphorodithioate] in surface runoff water from soil grown to sugarcane. This study focused on the fate of applied herbicides (atrazine and metribuzin). Results on the fate of azinphosmethyl, its rate of disappearance on leaves and soils, and runoff losses have been reported earlier (Granovsky et al., 1996). The experimental plots and field instrumentation layout are described in detail in Granovsky et al. (1996) and Bengtson et al. (1998). Briefly, the study was performed at the St. Gabriel Research Station of the Louisiana State University AgCenter, located about 16 km south of Baton Rouge, LA. The experimental site was approximately 1.5 ha and the soil was classified as Commerce silt loam. The land was rowed where six plots (three treatments x two replications) running east to west were established. Each plot consisted of nine 150-m-long rows with 1.82-m row spacing. Two additional rows served as levees separating adjacent plots and provided drainage from areas outside the plots. At the lowest corner of each plot, a sump made of corrugated iron was dug (1.5 m in diameter, 1.8 m deep) to collect runoff water. A float-controlled electric pump was installed in each sump to discharge the runoff through a water meter and into a drainage ditch. An automatic water sampler was used to collect runoff samples. All plots were planted with sugarcane variety CP70-321. The treatments consisted of broadcast versus band application of herbicides. Specifically, atrazine as well as metribuzin were applied on top of the sugarcane rows with a 0.6- or 0.9-m band width, or uniform or broadcast application where the entire area was treated. Each herbicide was tank-mixed and applied with a spray-rig with adjustable nozzles 45 cm above the ground surface. Atrazine (4 L) was applied as a winter herbicide on 6 Jan. 1994 and 20 Dec. 1994. The amounts (active ingredient) applied were 2.24, 1.12, and 0.74 kg ha-1 for the broadcast, 0.9-m band, and 0.6-m band treatments, respectively. Metribuzin was applied each spring on 29 Mar. 1993, 18 Mar. 1994, and 2 May 1995 as a postemergence herbicide, and was applied with the spray rig in a similar manner to atrazine. The amounts of metribuzin applied were 2.01, 1.01, and 0.67 kg ha-1, for the broadcast, 0.9-m band, and 0.6-m band treatments, respectively.

After each herbicide application (atrazine or metribuzin), composite samples of surface soil (to a 25-mm depth) were taken along one of the five center rows for each of the six plots. The rows within each plot were alternated randomly between the sampling dates. Sampling was performed daily for the first few days, and 5 to 7 d apart thereafter. Runoff water was automatically sampled following rainfall events that caused sufficient runoff to occur. All runoff water samples were stored in glass bottles at 4°C until laboratory analysis for nutrients and pesticides. In the laboratory, extraction of pesticides from soil and runoff water was as follows. Extractions from field (moist) soil were performed with 0.01 M NaCl methanol and water solution (4:1 v/v), shaking for 24 h, centrifuging, decanting, and removing of water with anhydrous sodium sulfate. The extracts were subsequently evaporated and transferred in hexane to 2.0-mL vials. Runoff water samples (250 mL) were extracted with dichloromethane in a separatory funnel. The remaining steps were similar to those for soil extractions. The herbicides were measured with a 5890A Series II gas chromatograph (Hewlett-Packard, Palo Alto, CA) with He as carrier gas and 63Ni electron capture detector. A coiled, fused PAS-1701 silica capillary column (25 m in length with a 0.32-mm i.d.) was used with a column flow of 1 mL min-1. Oven temperature was programmed to rise from 80 to 190°C at 30°C min-1 and then to 260°C at a rate of 3.6°C min-1. The injection and detector temperatures were maintained at 250 and 300°C, respectively. The retention times for atrazine and metribuzin were 7.96 ± 0.02 and 9.70 ± 0.02 min, respectively. The lower limits of detection were 0.05 and 0.10 µg L-1 for runoff water for atrazine and metribuzin, respectively.

Adsorption and Desorption
Batch experiments were conducted to determine adsorption and desorption isotherms for atrazine and metribuzin by the Commerce soil. The technique used here is kinetic batch type and is described in previous studies (Ma et al., 1993; Ma and Selim, 1996). Seven initial metribuzin concentrations (C0 = 1.39, 5.40, 10.29, 30.29, 60.25, 80.23, and 100.2 mg L-1) were prepared in 0.005 M CaCl2 solution and spiked with 14C-ring labeled metribuzin. In a similar fashion, for atrazine six initial concentrations (C0 = 1.80, 2.5, 5.4, 10.3, 20.2, and 30.0 mg L-1) were prepared in 0.005 M CaCl2 solution and spiked with 14C-ring labeled atrazine. Batch experiments were initiated by mixing 10 g of air-dry soil with 30 mL of metribuzin or atrazine solution in a 40-mL Teflon tube. Triplicates were used for each input concentration and the amount of pesticide adsorbed was taken as the average of the three replicates. The mixtures were kept shaking and centrifuged at 500 x g for 10 min for each specific reaction time before sampling. A 0.5-mL aliquot was sampled from the supernatant at reaction times of 2, 6, 12, 24, 48, 96, 192, and 384 h. After sampling, the slurry was agitated with a vortex mixer and returned to a shaker. The collected samples were analyzed with a Tri-Carb liquid scintillation ß counter (Packard-2100 TR) by mixing the 0.5-mL aliquot with 5 mL Packard Ultima Gold cocktail (PerkinElmer, Wellesley, MA) for 10 min on the liquid scintillation counter. The radioactivity was recorded as counts per minute, and the amount of pesticide retained by the soil at each reaction time was calculated from the difference in concentration of the supernatant and that of the initial solution.

The extent of release or desorption was also quantified with the batch method described above. For each initial concentration (C0), desorption commenced immediately after the last adsorption time step (384 h). Sequential or successive dilutions of the slurries were carried to induce pesticide release or desorption. Each desorption step was conducted by replacing the supernatant with pesticide-free 0.005 M CaCl2 background solution and shaking for 24 h. Six desorption steps were performed with a total desorption time of six days. Atrazine or metribuzin concentration in solution following each desorption step was also analyzed with liquid scintillation counting.

Data Analysis
Adsorption and desorption isotherms were used to estimate the distribution coefficients and Freundlich parameters based on nonlinear regression.


   RESULTS AND DISCUSSION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Adsorption–Desorption
Metribuzin and atrazine concentrations in the soil solution are plotted against reaction time for the various initial concentrations (C0) in Fig. 1 and 2 . The change of concentrations versus time clearly indicates that metribuzin adsorption kinetics was lacking. In fact, metribuzin adsorption reached near equilibrium within a few hours as illustrated by the lack of decrease in pesticide concentration versus time. Lack of adsorption kinetics can be further illustrated by the isotherms shown in Fig. 3 for a wide range of retention times (2–384 h). Adsorption isotherms, which represent the amount sorbed versus concentration in the soil solution, are commonly used to quantify the affinity of sorption and are often described by Freundlich- or Langmuir-type models. In this study, adsorption results were described based on the Freundlich approach:

[1]
where S is the amount sorbed (mg kg-1 soil), C is concentration in the liquid phase (mg L-1), Kf is partitioning coefficient (L kg-1), and N is a dimensionless parameter commonly less than unity. For cases where N = 1, we have the linear form:

[2]
where the parameter Kd (L kg-1) is the solute distribution coefficient, which is commonly reported in the literature. Best-fit Kd values for metribuzin ranged from 1.12 to 1.25 L kg-1 for 2 and 384 h of retention time, respectively (see Table 1 and Fig. 4) . Respective values for the Freundlich coefficient Kf also exhibited a lack of kinetics with the nonlinear parameter N close to unity (0.892–0.975). Based on such small Kd values, metribuzin is best regarded as a weakly sorbed herbicide. Ma and Selim (1996) showed a lack of kinetic retention and an extremely small value (Kd = 0.23 cm3 g-1) for a Cecil (fine, kaolinitic, thermic Typic Kanhapludult) kaolinitic soil and Kd of 0.74 L kg-1 for a Sharkey montmorillonitic clay soil. Savage (1976) reported Kd values for metribuzin in 16 soils from the lower alluvial flood plain of the Mississippi River. The reported Kd values are within a similar range, indicating low retention of metribuzin and thus high potential mobility in soils. It was also shown that metribuzin was well correlated with clay and organic matter contents (Savage, 1976; Harper, 1988). Johnson and Pepperman (1995) reported Kd values of 0.24 to 0.6 L kg-1 for four alluvial surface soils from central Louisiana with lower Kd values (0.09–0.47) for the subsurface soil layers. In a leaching experiment, Peter and Weber (1985) found that metribuzin was considerably more mobile in soil than atrazine. Other studies have shown the susceptibility of metribuzin to leaching in soils (Southwick et al., 1995; Starr and Glotfelty, 1990).



View larger version (21K):
[in this window]
[in a new window]
  		Fig. 1. Metribuzin concentration in soil solution versus time during adsorption for Commerce soil with different initial concentrations (C0). Solid curves are predictions using the multireaction model (MRM).

 


View larger version (24K):
[in this window]
[in a new window]
  		Fig. 2. Atrazine concentration in soil solution versus time during adsorption for Commerce soil with different initial concentrations (C0). Solid curves are predictions using the multireaction model (MRM).

 


View larger version (22K):
[in this window]
[in a new window]
  		Fig. 3. Adsorption isotherms for metribuzin by Commerce soil for 2-, 24-, and 384-h reaction times.

 

View this table:
[in this window]
[in a new window]
  		Table 1. Estimated linear and Freundlich model parameters and their standard errors (SE) of atrazine and metribuzin adsorption by Commerce soil at different reaction times.

 


View larger version (16K):
[in this window]
[in a new window]
  		Fig. 4. Distribution coefficient (Kd) versus reaction time for metribuzin and atrazine by Commerce soil.

 
Atrazine adsorption in Commerce soil exhibited kinetic retention behavior as indicated by the adsorption isotherms shown in Fig. 5 . Estimates for the parameters Kd, Kf, and N versus time were also determined to assess the extent of retention of atrazine with time (see Table 1). The Kd values for atrazine increased from 2.10 to 2.35 L kg-1 after 24 and 384 h of reaction time, respectively. The dependence of Kd versus retention time shown in Fig. 4 clearly indicates higher retention as well as stronger kinetic retention for atrazine by this Commerce soil in comparison with that for metribuzin. Estimated values for Kf exhibited similar kinetic retention with that for Kd as indicated by the results shown in Table 1. Here, Kf increased from 2.55 to 3.81 L kg-1 for 2 and 384 h of retention, respectively. The estimated nonlinear parameter N was significantly different from unity and ranged from 0.821 to 0.888. No consistent trend of N values versus time was observed, however. Ma et al. (1993) reported an N value of 0.877 for a Sharkey clay after 24 h of adsorption. However, Ma et al. (1993) reported higher Kf values and kinetic behavior of atrazine for a Sharkey clay soil compared with the Commerce soil.



View larger version (25K):
[in this window]
[in a new window]
  		Fig. 5. Adsorption isotherms for atrazine by Commerce soil for 2-, 24-, and 384-h reaction times. Solid curves are predictions using a linear model.

 
Adsorption–desorption results for metribuzin and atrazine are presented as isotherms in the traditional manner in Fig. 6 and 7 , respectively. These isotherms clearly indicate hysteresis for both pesticides by Commerce soil. Based on the discrepancy between the adsorption isotherm (after 384 h) and subsequent desorption or release isotherms, it is obvious that hysteretic behavior was less extensive for metribuzin compared with that for atrazine. This is consistent with the observation in this study of the kinetic behavior of atrazine in comparison with metribuzin in Commerce soil. Lack of equilibrium conditions may also be responsible for observed desorption hysteresis. In fact, Xue and Selim (1995) postulated that hysteresis was perhaps due to irreversible adsorption by the soil matrix. Atrazine as well as metribuzin may be retained by heterogeneous sites having a wide range of binding energies. At trace concentrations, binding may be irreversible and the amount of nondesorbable pesticide almost always increased with time. It was reported that hysteresis may be due to hydrophobic bonding of the chemical to soil organic matter (Pusino et al., 1994). This lack of complete recovery of soil-applied pesticides has been reported by many others (e.g., Bowman and Sans, 1985; Pignatello and Huang, 1991). Based on higher Kd values and hyteresis, it can be concluded that atrazine is less susceptible to transport in this soil compared with metribuzin.



View larger version (27K):
[in this window]
[in a new window]
  		Fig. 6. Adsorption and desorption isotherms for metribuzin by Commerce soil and different initial concentrations (C0).

 


View larger version (28K):
[in this window]
[in a new window]
  		Fig. 7. Adsorption and desorption isotherms for atrazine by Commerce soil and different initial concentrations (C0).

 
Multireaction Model
The retention parameters given in Table 1 for Kf and Kd indicate time-dependent behavior of atrazine and metribuzin behavior by Commerce soil (see also Fig. 4). Therefore, the use of a kinetic model to describe the time-dependent behavior for both herbicides is justified. The simulation data represented by the solid curve shown in Fig. 1 and 2 are an attempt to describe the retention kinetics for atrazine and metribuzin over the initial concentrations (C0) used in the batch experiments. The model used was the multireaction model (MRM) described by Xue and Selim (1995). This multipurpose model accounts for equilibrium- and kinetic-type retention reactions of reactive solutes in soils. The model version chosen in this analysis can be presented by the following formulation:

[3]

[4]

[5]

[6]
where C is the concentration in soil solution (mg L-1), S is the total amount sorbed or retained by the soil matrix (mg kg-1), Se is assumed to represent the amount retained on equilibrium-type sites (mg kg-1) and has a low binding energy, Sk is the amount retained on kinetic-type sites (mg kg-1) through strong interactions with the soil matrix, and Si represents the amount retained irreversibly (mg kg-1). The coefficient Ke is an equilibrium constant (dimensionless) associated with instantaneous reactions, whereas k1 and k2 (h-1) are the forward and backward reaction rate coefficients associated with the kinetic-type sites, respectively. The parameter k3 (h-1) is the irreversible rate coefficient associated with the kinetic sites. The parameter n is the reaction orders (dimensionless) associated with Se and Sk, {theta} is the soil water content (cm3 cm-3), {rho} is the soil bulk density (g cm-3), and t is time (h).

To obtain the simulation shown in Fig. 1 and 2, the multireaction model (MRM) was used along with a nonlinear least-squares optimization scheme, which provided a best-fit of the model to the experimental data. It was assumed that if the model is incapable of describing measured concentration results, the model is an inaccurate representation of the retention mechanisms. For metribuzin, the goodness of fit as measured by r2 was 0.999 (Fig. 1), whereas an r2 of 0.997 was obtained for atrazine (Fig. 2). To evaluate the performance of the MRM model, a model diagnostic analysis was performed by plotting experimentally measured results versus model predictions for the entire data set. Simple linear regression analysis of measured versus predicted values was also performed, separately, for metribuzin and atrazine data sets. Then a t test was performed on the intercepts and slopes of the regression lines to ascertain whether they are significantly different from 0 and 1, respectively. Based on regression analysis, the slope and intercept of the regression line were 1.00 ± 0.067 and -0.012 ± 0.304 for metribuzin, whereas the respective values were 1.00 ± 0.014 and -0.017 ± 0.0122 for atrazine (figures not shown). Based on the t test, the p values for metribuzin were 0.933 and 0.963 for the intercept and slope, respectively. The respective p values for atrazine were 0.803 and 0.989. Therefore, based on the t test, for both metribuzin and atrazine, the assumption that the slope and intercept of the regression lines is not significantly different from 1 and 0, respectively, is not rejected, and thus it can be concluded that MRM provides adequate prediction of the data.

Parameter estimates of the MRM models, which provided the best-fit of metribuzin results, were 0.241, 0.620, and 0.000142 h-1 for k1, k2, and k3, with standard errors of 0.0399, 0.050, and 0.000118 h-1, respectively. The estimate for the parameter n was 0.884 with a standard error of 0.0336. For the atrazine simulations shown in Fig. 2, the parameter estimates for n, k1, k2, and k3 were 0.816, 0.533, 0.609, and 0.00059 h-1 and their standard errors were 0.0269, 0.0582, 0.0489, and 0.000115 h-1, respectively. These estimates represent "overall" model parameters for the entire adsorption data set; that is, for all input concentrations (C0) for metribuzin as well as atrazine. Although a comparison of parameter estimates based on MRM and those based on equilibrium (linear and Freundlich) models is not explicitly applicable, several consistent trends need to be emphasized. First, the magnitude of the rate coefficients (k1 and k2) compared with k3 depicts the fast kinetic or rapid reaction as indicated by the initially fast adsorption that occurred within the first few hours. For large times (t -> {infty}), values for Kf can be approximated based on [({theta}/{rho})k1/k2]. Second, for both herbicides, values of the Freundlich N of Eq. [1] were consistent with the nonlinear parameter n of Eq. [5] of MRM (see Table 1).

Based on model simulations, metribuzin and atrazine concentrations versus time were well described by the MRM model (Fig. 1 and 2). Moreover, for each herbicide, one set of model parameters is applicable for the entire concentration range, that is, for all C0 values. For a similar data set of atrazine retention by a Sharkey soil, Ma and Selim (1996) estimated an individual set of parameters for each input concentration (C0). Ma and Selim (1996) concluded that an "overall" set of parameters for all C0 values provided equally good prediction of atrazine retention when compared with individually fitted parameters. Modeling efforts based on multireaction mechanisms for metribuzin were not found in a literature search.

Runoff Losses
During the three years (1993–1995) of the study, the average annual rainfall was 1449 mm, which was close to the normal of 1445 mm. The average annual surface runoff was 689, 684, and 687 mm (27.1, 26.9, and 27.1 in) for the broadcast, 0.9-m band, and 0.6-m band treatments, respectively, with no significant differences among the three treatments (p = 0.258). Moreover, for all treatments, there were no significant differences among the 1993, 1994, and 1995 runoff results (p = 0.994). Runoff results during the fallow year (1996) are reported in Bengtson et al. (1998).

Figures 8 and 9 show averaged atrazine runoff concentration for the 1994 and 1995 growing seasons. Amounts of atrazine applied were 2.2, 1.1, and 0.7 kg ha-1 for the broadcast, 0.9-m band, and 0.6-m band treatments, respectively. However, the intensity of herbicide application was maintained constant (amount per area of coverage) for all three treatments. Atrazine concentration levels in the runoff ranged from 600 to 700 µg L-1, 6 d following applications for the broadcast treatment in 1994. For the 0.9-m band treatment, the concentration level was drastically reduced to a range of 120 to 140 µg L-1 and for the 0.6-m band treatment the concentrations did not exceed 80 µg L-1. Therefore, broadcast application of atrazine resulted in as much as five to eight times higher initial concentrations of atrazine in the runoff when compared with that for band applications. Three months following herbicide applications, atrazine concentrations detected in surface runoff were drastically reduced to 20 to 30 µg L-1 for all three treatments.



View larger version (23K):
[in this window]
[in a new window]
  		Fig. 8. Atrazine concentrations in water runoff from the broadcast, 0.9-m band, and 0.6-m band treatments during 1994 (average of two replications).

 


View larger version (20K):
[in this window]
[in a new window]
  		Fig. 9. Atrazine concentrations in water runoff from the broadcast, 0.9-m band, and 0.6-m band treatments during 1995 (average of two replications).

 
In 1995, atrazine was detected in the runoff 8 and 11 d following application. The concentration ranged from 12 to 46 µg L-1 for the 0.6-m band treatment and 100 to 480 µg L-1 for the broadcast treatment. One month later, atrazine concentrations decreased considerably for all treatments and reached 15 to 20 µg L-1 for the 0.6-m band treatment and 50 to 75 µg L-1 for the broadcast. Differences among treatments continued to diminish in a similar manner to the 1994 results, although higher concentrations were consistently observed for plots having received larger applications (broadcast > 0.9 m-band > 0.6 m-band). Two months after application, atrazine concentrations in the runoff waters were further reduced to 4 to 10 µg L-1 for all plots, without noticeable difference among the different treatments.

From measured concentrations and runoff volumes, the total amount of herbicide losses in the runoff from each plot treatment were calculated. During the 1994 growing season, the total amounts of atrazine losses in the runoff were 53, 92, and 249 g ha-1 for the 0.6 m-band, 0.9 m-band, and broadcast treatments, respectively. These runoff losses represent 7.2 to 11.1% of applied atrazine (Table 2). Thus, it can be concluded that a significant reduction of the total atrazine losses in runoff water was achieved for band application in comparison with full broadcast. In fact, in 1994 the amounts of losses were reduced by 63% for the 0.9-m band and 78% for the 0.6-m band treatments when compared with full broadcast. For 1995, the amount of atrazine losses were considerably less than that for the 1994 growing season and ranged from 1.91 to 2.88% of that applied. Southwick et al. (1992) reported total losses of 1.6 to 2.6% of atrazine applied at 4.48 kg ha-1 to sugarcane grown on a Sharkey clay soil. Essentially, all of these losses occurred in the first 21 d with accumulated rainfall of 203 mm. Runoff losses of atrazine as high as 7.3% of the amount applied after a 6-mm rainfall were reported by White et al. (1967). Moreover, maximum losses by run-off are typically the result of a single runoff event (Wauchope, 1978), with the highest concentrations measured when heavy rainfall events occur close to application.


View this table:
[in this window]
[in a new window]
  		Table 2. Atrazine and metribuzin losses in runoff water during 1994 and 1995.

 
Metribuzin concentrations in the runoff following spring applications in 1994 and 1995 are shown in Fig. 10 and 11 . For 1994, two weeks of no rainfall followed the 18 March metribuzin application. In fact, runoff amounts after the 27 March and 4 April rainfalls (15 and 25 mm, respectively) were negligible. Subsequent rainfall events on 12 and 15 April resulted in significant runoff waters. Metribuzin concentrations following these events ranged from 10 to 30 µg L-1. Some two months later, metribuzin concentrations decreased drastically to levels below 1 µg L-1. In 1995, the first runoff event occurred 6 d following metribuzin application (8 May) resulting in high initial runoff concentrations. The range was 250 to 450 µg L-1 from the broadcast plots and 12 to 46 µg L-1 from the 0.6-m band. Only one runoff sample from the broadcast plot was collected 17 d following application where the concentration of metribuzin was 80 µg L-1. This concentration compares well with that for 1994 when 60 µg L-1 was observed from the broadcast plot 28 d after application. Furthermore, two months later (29 June), concentrations in the runoff were less than 4 µg L-1 for all treatments. In fact, metribuzin concentrations were less than 1 µg L-1 in runoff waters five months following application. Such findings are consistent with those for atrazine. As shown in Table 2, cumulative amounts of metribuzin losses in runoff water were highest for full broadcast (0.3–7.4% of the amount applied) and lowest for the 0.6-m band treatment (0.3–3.4%). Thus, it can be concluded that significant reduction of metribuzin losses was achieved when band rather than broadcast applications were implemented (P < 0.05, LSD).



View larger version (27K):
[in this window]
[in a new window]
  		Fig. 10. Metribuzin concentrations in water runoff from the broadcast, 0.9-m band, and 0.6-m band treatments during 1994 (average of two replications).

 


View larger version (24K):
[in this window]
[in a new window]
  		Fig. 11. Metribuzin concentrations in water runoff from the broadcast, 0.9-m band, and 0.6-m band treatments during 1995 (average of two replications).

 
Several investigations measured high initial concentrations of herbicides in runoff waters from soils under various management practices (Southwick et al., 1992, 1997; Isensee and Sadeghi, 1999). For example, Isensee and Sadeghi (1993) quantified concentrations and runoff amounts of atrazine, alachlor, and cyanzine herbicides from no-till (NT) and conventional tillage (CT) 0.25- to 0.50-ha plots. All herbicides were tank-mixed and applied (broadcast) uniformly across the plots one day after planting. They found that atrazine concentrations were 2 to 10 times higher in runoff water from the NT compared with the CT plots. Highest atrazine concentrations reported were 3 d following applications and ranged from 2061 to 2733 µg L-1 for NT and 773 to 989 µg L-1 for CT. Such concentrations are considerably higher than those measured in this experiment. Isensee and Sadeghi (1993) also reported that the values decreased to 36 µg L-1 for NT and 3 to 4 µg L-1 for CT some 34 d following applications. Such values were considerably lower than those measured in these experiments, indicative of perhaps higher degradation and runoff. Results from the 0.6- and 0.9-m band applications were consistently lower than those reported for broadcast applications by Isensee and Sadeghi (1993).

Disappearance from Surface Soil
Concentrations of extractable atrazine and metribuzin from the surface soil (0–2.5 cm deep) exhibited a sharp decrease with time following application for both growing seasons as shown in Fig. 12 and 13 . Such relations are often referred to as dissipation or disappearance curves (see Southwick et al., 1995) and are influenced by the retention of herbicide with the soil matrix, chemical and biological transformations, volatilization, downward movement, and runoff. Maximum extractable atrazine concentration occurred 3 to 4 d following application for all three treatments. Lowest herbicide concentrations were consistently measured for the 0.6-m band and to a lesser extent the 0.9-m band treatment when compared with full broadcast. This is perhaps due to possible sampling of surface soil on top of the sugarcane rows outside the treated 0.6-m band area. Because the only major difference among the treatments is herbicide band width, a similar pattern of disappearance for each herbicide from the surface soil is expected.



View larger version (26K):
[in this window]
[in a new window]
  		Fig. 12. Extractable atrazine concentrations from the surface soil from all treatments during 1994 (top) and 1995 (bottom). Solid curves are predictions using Model 3.

 


View larger version (27K):
[in this window]
[in a new window]
  		Fig. 13. Extractable metribuzin concentrations from the surface soil from all treatments during 1994 (top) and 1995 (bottom). Solid curves are predictions using Model 3.

 
Because of the extensive scattering of extractable herbicide concentrations from surface soil following application, efforts to quantify the rate of decay or disappearance were limited. Such data scattering is not easily understood and was perhaps due to soil heterogeneity and wind drift during application, among others. Nonlinear regression analysis was used to provide estimates for rates of disappearance of applied atrazine and metribuzin from the surface soil over the two growing seasons (see Fig. 12 and 13). The assumption of first-order or exponential decay was used here. Specifically, SAS procedure PROC NLIN was used (SAS Institute, 1999). Three model versions were tested to assess the effect of the different treatments (band versus broadcast). Model 1 is a general-purpose decay model, which can be expressed as:

[7]
where S (mg kg-1) is herbicide concentration at time t (d), and {alpha} and ß are dummy variables indicating the different treatments. Here {alpha} = 0 and ß = 0 for the 0.9-m band treatment, {alpha} = 1 and ß = 0 for the broadcast treatment, and {alpha} = 0 and ß = 1 for the 0.6-m band treatment. Moreover, S0, S1, and S2 (mg kg-1) are herbicide concentrations at t = 0 associated with the three different treatments. Specifically, S0, S0 + S1, and S0 + S2 represent initial concentrations for the 0.9-m band, broadcast, and 0.6-m band treatments, respectively. Therefore, estimates of S1 or S2 that are significantly different from zero signify significant difference between the treatments. The role of the rate coefficients {lambda}, {lambda}1, and {lambda}2 (d-1) in the model is incorporated in the same way.

Model 2 is a simplification of Model 1 except that an overall rate coefficient {lambda} is assumed. This assumption is based on the fact that differences between treatments, if any, should be only due to differences in herbicide amount on the soil (S0) and not their rate of disappearance. Therefore, Model 2 reads as follows:

[8]

Finally, Model 3 is the simplest model where an overall first-order decay is assumed to describe an entire data set such that:

[9]

An analysis of variance (ANOVA) table for the regression analysis (not shown) suggests a first-order approach provided a good description of the metribuzin and atrazine dissipation data based on all three models. For the 1994 metribuzin data, overall dissipation rates {lambda} of 0.0898 and 0.0840 d-1 were estimated based on Models 2 and 3, respectively (see Table 3 and 4). These values correspond to half-lives (t1/2) of 7.71 and 8.25 d, respectively. Based on Model 1 estimates, significant differences in {lambda} between treatments were obtained. The corresponding t1/2 values were 5.80, 10.1, and 11.3 d for the broadcast, 0.6-m band, and 0.9-m band treatments, respectively. Differences of {lambda} values for the different treatments illustrate the extensive data scattering of field results as discussed above. For the 1995 metribuzin data, no significant differences of dissipation among the three treatments were observed, however. The t1/2 values based on Model 1 were 5.8, 6.2, and 5.8 d for the broadcast, 0.6-m band, and 0.9-m band treatments, respectively. Such values were comparable with those based on Models 2 and 3, where overall t1/2 estimates were 5.95 and 5.97 d, respectively. Such a finding suggests that the use of an overall {lambda} (or t1/2) is adequate in describing the surface soil data of this study. The curves shown in Fig. 12 and 13 are based on Model 3, where overall {lambda} values (Model 3) were used for metribuzin and atrazine disappearance, respectively.


View this table:
[in this window]
[in a new window]
  		Table 3. Parameter estimates of the first-order model for atrazine dissipation from the surface soil during 1994 and 1995.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Parameter estimates of the first-order model for metribuzin dissipation from the surface soil during 1994 and 1995.

 
For atrazine, estimates for {lambda} (or t1/2) were consistently higher than for metribuzin (see Table 4). For example, for 1995, overall t1/2 estimates for atrazine were 24.2 and 24.6 d, based on Models 2 and 3, respectively. For 1994, smaller overall t1/2 for atrazine was obtained (13.5 and 13.54 d, respectively). The faster atrazine disappearance during 1995 compared with 1994 cannot be explained. Southwick et al. (1992) estimated atrazine half-lives in a Sharkey soil of 24 and 102 d in the 1989 and 1990 growing seasons, respectively. They postulate that such differences may be due to the influence of soil temperature as well as moisture content of the soil. In a later study, Southwick et al. (1995) reported half-lives for atrazine of 14.4 and 21 d for the winter of 1991 and summer of 1992, respectively. Such results are consistent with the estimated half-lives from this study.

Simple regression analysis was performed on the measured values versus predictions based on the first-order decay model for metribuzin and atrazine given in Fig. 12 and 13, respectively. This diagnostic analysis was performed to evaluate the performance of the first-order model in describing field results by plotting measured field results versus model predictions for each data set separately (1:1 plots). Then, a t test was performed on the intercepts and slopes of the regression lines. Based on the t test, the p values for metribuzin and atrazine ranged from 0.163 to 0.501. Therefore, the respective slopes and intercepts of the regression lines were not significantly different from 1 and 0, respectively, and thus conclude that the first-order decay model provides adequate prediction of field results for both metribuzin and atrazine. The only exception was for the 1995 atrazine model results where p values of 0.102 and 0.011 for the slope and intercept of the regression lines were obtained, respectively. An explanation for model failure for this data set was not found.


   CONCLUSIONS
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
In this study, the dynamics of metribuzin and atrazine in surface soils and runoff waters in sugarcane fields were investigated. Metribuzin was applied in the spring as a postemergence herbicide and atrazine was applied following winter harvest. Both herbicides were applied on top of the sugarcane rows as 0.6-m band width, 0.9-m band width, or broadcast applications where the entire area was treated. Based on effluent concentrations, it can be concluded that a significant reduction of the total atrazine and metribuzin losses in runoff water was achieved for band application in comparison with full broadcast. Maximum effluent concentrations were measured from the broadcast treatment and ranged from 600 to 1100 µg L-1 for atrazine and 250 to 450 µg L-1 for metribuzin. Atrazine runoff losses were highest for broadcast treatment (2.8–11% of that applied) and lowest for the 0.6-m band treatment (1.9–7.6%) with a similar trend for metribuzin losses. Measured extractable herbicides from the surface soil exhibited a sharp decrease with time and were well described with a simple first-order decay model. For atrazine, estimates for the rate of decay ({lambda}) were higher than for metribuzin.

Retention (adsorption - desorption) results for atrazine and metribuzin based on laboratory measurements were consistent with field observations. Adsorption was performed with a kinetic batch method for a wide range of concentrations. The distribution coefficients (Kd) for atrazine exhibited stronger retention over time in comparison with metribuzin on Commerce soil. A mechanistic (kinetic-equilibrium) multireaction model provided good prediction of the kinetic retention behavior of atrazine and metribuzin for a wide range of concentrations. Atrazine and metribuzin desorption were quantified with successive dilutions following adsorption. Desorption results indicated discrepancies between adsorption and desorption isotherms, which is indicative of slow release. Moreover, the extent of hysteresis was more pronounced for atrazine compared with metribuzin.


   ACKNOWLEDGMENTS
 
The author wishes to thank Bayer Corporation, Agricultural Division, Kansas City, for providing the radio-labeled herbicides used in this study, and K. Wang, H. Zhu, and L. Zhou for their assistance. This study was funded in part by the Louisiana Department of Environmental Quality (Contract 24400-93-32), Nonpoint Source Program, Baton Rouge, LA.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Approved by the Director of the Louisiana Agricultural Experiment Station as Manuscript no. 02-09-0488.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 


Related articles in JEQ:

This Issue in Journal of Environmental Quality

JEQ 2003 32: 745-750. [Full Text]  



This article has been cited by other articles:


Home page
 Soil Sci.Home page
L. Ma and H. M. Selim
Predicting Pesticide Transport in Mulch-Amended Soils: A Two-Compartment Model
Soil Sci. Soc. Am. J., March 1, 2005; 69(2): 318 - 327.
[Abstract] [Full Text] [PDF]

Home page
 J. Environ. Qual.Home page
H. M. Selim and H. Zhu
Atrazine Sorption-Desorption Hysteresis by Sugarcane Mulch Residue
J. Environ. Qual., January 1, 2005; 34(1): 325 - 335.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Related articles in JEQ
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via ISI Web of Science (8)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Selim, H. M.
Right arrow Search for Related Content
<