In situ measurements and LEACHM predictions of the transport and fate of nonreactive tracers and dicamba in a silt loam Montana soil by Robert John Pearson A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Soils Montana State University © Copyright by Robert John Pearson (1994) Abstract: Groundwater contamination by pesticides has been documented in thirty-eight states. Regionally, seven pesticides have been detected in Montana groundwater. Solute transport models such as LEACHM, PRZM and CMLS have become increasingly important tools for predicting the fate of chemicals applied to soils. Computer simulation models based on analytical solutions of the convective dispersion equation (CDE) have been developed to screen pesticides for their leaching potential, to screen soil mapping units for pesticide leaching potential, and to predict the fate of chemicals at specific locations. Despite growing interest in the application of solute transport models such as LEACHM for predicting chemical movement in soils, there have been relatively few studies conducted to test model performance under field conditions. Groundwater tracers used in validation studies differ in their suitability for studies of solute transport in soils. Bromide has long been accepted as a good soil water tracer, however Br has recently been shown to be absorbed by plants. Under these conditions Br mass would not remain constant in the soil profile. In situ transport studies involving fluorobenzoate tracers, bromide and 14C ring-labelled dicamba (3,6-dichloro-2-methoxybenzoic acid) were performed to evaluate tracer suitability, pesticide behavior and to validate the Leaching Estimation and Chemistry Model (LEACHM) under field conditions in a silt loam Montana soil. Prior to field work, a laboratory study was conducted to develop methodology for the analysis of fluorobenzoate tracers using an ion chromatograph (IC) equipped with an electrical conductivity detector. Two successive field studies were then conducted at a field site near Manhattan, MT during the summers of 1991 and 1992. Two treatments, fallow and crop (Hordeum vu/gare L.), over three different water regimes were established perpendicular to a single line source irrigation system. The IC methodology developed allowed for expedient analysis of fluorobenzoate tracers and Br Results of tracer evaluation indicated that significant plant uptake of Br occurred. Leaching of tracers and dicamba occurred to depths greater than 1.0 m. Dicamba's primary metabolite- 3,6-dichlorosalicylic acid (DCSA) was less mobile and persisted longer than dicamba LEACHM performed best under 1991 field conditions relative to 1992 field conditions. Evidence of preferential flow was greatest in the 1992 field data where decaying plant roots from the previous study possibly contributed to the discrepancies between predicted and observed data.  IN SITU MEASUREMENTS AND LEACHM PREDICTIONS OF THE TRANSPORT AND FATE OF NONREACTIVE TRACERS AND DICAMBA IN A SILT LOAM MONTANA SOIL by Robert John Pearson A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Soils MONTANA STATE UNIVERSITY Bozeman, Montana June 1994 ii APPROVAL of a thesis submitted by Robert John Pearson This thesis has been read by each member of the thesis committee and has been found to be satisfactory regarding content, English usage, format, citations, bibliographic style, and consistency, and is ready for submission to the College of Graduate Studies. /Tzl / f f f Date zy 6 ^mmChairperson, Graduate Cc^jnmittee Approved for the Major Department Date Approved for the College of Graduate Studies --------- Graduate DeanDate iii STATEMENT OF PERMISSION TO USE In presenting this thesis in partial fulfillment of the requirements for a master's degree at Montana State University, I agree that the Library shall make it available to borrowers under rules of the Library. If I have indicated my intention to copyright this thesis by including a copyright notice page, copying is allowable only for scholarly purposes, consistent with "fair use" as prescribed in the U.S. Copyright Law. Requests for permission for extended quotation from or reproduction of this thesis in whole or in parts may be granted only by the copyright holder. Signature ~7 Date______ i* - f - f y V TABLE OF CONTENTS Page APPROVAL . ; ...........' ............................................................................. ................ii STATEMENT OF PERMISSION TO USE ................ ................'......................... ill V IT A ........................ iv TABLE OF CONTENTS........................................................................ .. . . v LIST OF TABLES ........................ vii LIST OF FIGURES ! ..................................................................... ' . . . . - .............. ix ABSTRACT ............................................. xi CHAPTER O N E ................................................ I INTRODUCTION........................................ I CHAPTER TWO ..................................................................................... 4 ANALYSIS OF FLUOROBENZOATE TRACERS BY ION CHROMATOGRAPHY............................................................. 4 Introduction.................................................................................. .. . 4 Materials and Methods ......................................................................6 Results and Discussion ........................... . . .................................7 CHAPTER THREE ..................................................................... .. ............. .............14 IN SITU MEASUREMENTS AND LEACHM PREDICTIONS OF THE TRANSPORT OF PENTAFLUOROBENZOIC ACID AND BROMIDE IN A MONTANA SILT LOAM SOIL ............. 14 Introduction................................ 14 Materials and Methods ............................................................ . 19 Sample Analysis . . ........................................................................... 22 LEACHM Simulations ........................................................... 25 Results and Discussion ................................ 27 Observed PFBA Transport ........................................ 27 Observed Br' Transport and Plant U p take ........................34 LEACHM Predicted Soil Water Contents ............. 38 LEACHM Predicted BTCs . ....... ..........................................42 Conclusions..................................................... ....................... . 43 ~i TABLE OF CONTENTS - Continued Page CHAPTER F O U R ........................................................................................................45 IN SITU MEASUREMENTS AND LEACHM PREDICTIONS OF THE. TRANSPORT AND FATE OF 2,6-DIFLUOROBENZOIC ACID AND DICAMBA IN A SILT LOAM MONTANA SOIL . . . ................ 45 Introduction................................................ 45 Materials and Methods .................................................................. 48 Field Study ................................ ........................r ................ 48 Sample Analysis..................................................................... 52 LEACHM Simulations ...........................................................53 Results and Discussion ........................................ 55 Observed Dicamba and 2,6-DFBA Breakthrough Curves55 LEACHM Predicted BTCS...................................................65 Dicamba and DCSA Concentrations in the Soil Profile .74 Conclusions..................... 80 SUMMARY ........................................................ 82 Laboratory and Field Studies ......................................................................83 LEACHM Performance .................................................................. 86 REFERENCES CITED ............. : . . . .................... ...........................................88 vi vii LIST OF TABLES Table Page 1. Retention times of mixed organic tracers and bromide standard (25 mg L"1) using IC analysis and corresponding retention times using HPLC analysis.................................................................................... 8 2. Profile characterization data of Brocko silt lo a m ..................... .. 21 3. Precipitation-irrigation and evaporation data used in LEACHM sim ulations..................................................... ............................ .. 24 4. Physical and chemical input parameters used in LEACHM sim ulations................................................................................ 26 5. . Moment analysis results based on selected observed and predicted PFBA and Br" breakthrough c u rv e s ..................................................... 33 6. Average Br concentrations in whole plant (Horeum vufgare L.) . tissue samples and percent of applied Br" measured as plant uptake at harvest ..................................... 37 7. . Observed and predicted water budget components from time of chemical application (t = 0) to final sampling event (t = 5 9 ) ............3 9 8. Physical and chemical input parameters used in LEACHM sim ulations............................................................................................. 50 9. Precipitation-irrigation and evaporation data used in LEACHM sim ulations.................. 51 10. Moment analysis results based on selected observed dicamba and 2,6-DFBA breakthrough curves ................................................................64 11. Observed and predicted water budget components from time of chemical application (t = 0) to final sampling event (t = 6 7 ) ............. 73 12. Observed and predicted 14C recoveries in in situ soil columns (120 cm) as a percent of total 14C applied as dicamba ................................. 78 viii LIST OF TABLES (continued) Table Page 13. Characterization of 14C fractions in selected soil samples at termination using solvent extraction, HPLC-radioisotope detection and total oxidation . . ...................................................... .......................... 79 ix LIST OF FIGURES Figure Page 1. Mixed ion chromatograph of 3 fluorobenzoates, Br and NO3 at 25.0 mg L'1 and Cl' at 1.0 mg L 1 using an output range detection of 30 /yS and an AS4A column ................................ .. ..................... .. 10 2. Relative peak areas of O-TFMBA, 2,6-DFBA, PFBA and Br' versus concentrations using ion chromatography. Data points represent means of three replicates; all standard error bars fall within sym b o ls ................................................................................ .. 12 3. Observed and simulated PFBA concentrations at 0 .36, 0.66 and 0.96 m depths under high, medium and low water regimes for fallow treatments. Vertical bars on symbols (observed data) indicate standard errors (n =4 ), where absent, bars fall within sym b o ls ................................................ .. . .................................................29 4. Observed and simulated PFBA and BR: concentrations at 0.36, . 0 .66 and 0.96 m depths under high, medium and low water regimes for crop treatments. Vertical bars on symbols (observed data) indicate standard, errors (n = 4),. where absent, bars fall within symbols ................................................................... .....................30 5. Observed and simulated soil volumetric water contents at 0.40, 0.60 and 1.0 m depths under, high, medium and low water regimes for fallow treatments, f indicates chemical application day. Vertical bars on symbols (observed data) indicate standard errors (n -2 ) , where absent, bars fall within symbols ..........., . . . 31 6. Observed and simulated soil volumetric water contents at 0.40, 0.60 and 1.0 m depths under high, medium and low water regimes for crop treatments. T indicates chemical application day. Vertical bars on symbols (observed data) indicate standard errors (n = 3), where absent, bars fall within sym bols...................................32 7. Observed and simulated daily surface evaporation from bare soil . 41 X LIST OF FIGURES - Continued Figure Page 8. Observed and simulated dicamba concentrations at 0.36, 0.66 and 0.96 m depths under high and medium water regimes for fallow treatments. Vertical bars on symbols (observed data) indicate standard errors (n.= 3), where absent, bars fall within sym b ols ........................................................................................................ 60 9. Observed and simulated dicamba concentrations at 0 .36, 0.66 , and 0.96 m depths under high and medium water regimes for crop treatments. Vertical bars on symbols (observed data) indicate standard errors (n =3 ), where absent, bars fall within symbols . . . 61 10. Observed and simulated 2,6-DFBA concentrations at 0.36, 0.66 and 0.96 m depths under high, medium and low water regimes for fallow treatments. Vertical bars on symbols (observed data) indicate standard errors (n = 3), where absent, bars fall within sym b o ls ........................................................................................................ 62 11. Observed and simulated 2,6-DFBA concentrations at 0 .36, 0.66 and 0.96 m depths under high, medium and low water regimes for crop treatments. Vertical bars on symbols (observed data) indicate standard errors (n = 3), where absent, bars fall within symbols . . . 63 12. Observed and simulated daily surface evaporation from bare soil . 70 13. Observed and simulated soil volumetric water contents at 0.40, 0.60 and 1.0 m depths under high, medium and low water regimes for fallow treatments, t indicates chemical application day. Vertical bars on symbols (observed data) indicate standard errors (n = 2), where absent, bars fall within symbols . . . . . . . . . 71 14. Observed and simulated soil volumetric water contents at 0.40, 0.60 and 1 .0 m depths under high, medium and low water regimes for crop treatments, t indicates chemical application day. Vertical bars on symbols (observed data) indicate standard errors (n = 3), where absent, bars fall within sym bols................................ . 7 2 15. Final observed and simulated dicamba and DCSA concentrations as a function of soil depth ......................................................................77 xi ABSTRACT Groundwater contamination by pesticides has been documented in thirty-eight states. Regionally, seven pesticides have been detected in Montana groundwater. Solute transport models such as LEACHM, PRZM and CMLS have become increasingly important tools for predicting the fate of chemicals applied to soils. Computer simulation models based on analytical solutions of the convective dispersion equation (CDE) have been developed to screen pesticides for their leaching potential, to screen soil mapping units for pesticide leaching potential, and to predict the fate of chemicals at specific locations. Despite growing interest in the application of solute transport models such as LEACHM for predicting chemical movement in soils, there have been relatively few studies conducted to test model performance under field conditions. Groundwater tracers used in validation studies differ in their suitability for studies of solute transport- in soils. Bromide has long been accepted as a good soil water tracer, however Br has recently been shown to be absorbed by plants. Under these conditions Br" mass would not remain constant in the soil profile. In situ transport studies involving fluorobenzoate tracers, bromide and 14C ring-labelled dicamba (3,6-dichloro-2-methoxybenzoic acid) were performed to evaluate tracer suitability, pesticide behavior and to validate the Leaching Estimation and Chemistry Model (LEACHM) under field conditions in a silt loam Montana soil. Prior to field work, a laboratory study was conducted to develop methodology for the analysis of fluorobenzoate tracers using an ion chromatograph (IC) equipped with an electrical conductivity detector. Two successive field studies were then conducted at a field site near Manhattan, MT during the summers of 1991 and 1992. Two treatments, fallow and crop (Hordeum vulgare L.), over three different water regimes were established perpendicular to a single line source irrigation system. The IC methodology developed allowed for expedient analysis of fluorobenzoate tracers and Br". Results of tracer evaluation indicated that significant plant uptake of Br" occurred. Leaching of tracers and dicamba occurred to depths greater than 1 .0 m. Dicamba's primary metabolite- 3,6-dichlorosalicylic acid (DCSA) was less mobile and persisted longer than dicamba. LEACHM performed best under 1991 field conditions relative to 1992 field conditions. Evidence of preferential flow was greatest in the 1992 field data where decaying plant roots from the previous study possibly contributed to the discrepancies between predicted and observed data. I CHAPTER ONE INTRODUCTION Well water monitoring surveys conducted in Montana have detected several pesticides including aldicarb, atrazine, 2,4-D, dicamba, MCPA, picloram and simazine (Deluca et al., 1989; Clark, 1990). Migration of these pesticides into groundwaters has occurred presumably through normal agricultural management practices. Nationally, over 70 pesticides have been found in groundwater in 38 states with 17 pesticides detected at concentrations above health advisory limits (Ritter, 1990; Parsons and Witt, 1988). The presence of pesticides in groundwaters has increased concern about the role of agricultural practices in the degradation of water quality, and has resulted in the development of numerous predictive computer models for evaluating the. movement of pesticides through soils. Computer models such as CMLS (Nofziger and Hornsby, 1987), PRZM (Carsel et al., 1984), GLEAMS (Leonard et al., 1987) and LEACHM (Wagenet and Hutson, 1989) have been used in a variety of applications including educational tools, screening pesticides for their leaching potential, screening mapping units for leaching potential and for predicting absolute concentrations of pesticides as a function of soil depth and time. Of the currently used deterministic transport models, LEACHM contains numerous subroutines 2 necessary to couple water flow, heat flow, solute sorption, solute degradation and plant water uptake. As such, LEACHM requires an exhaustive set pf input data on soil properties, solute properties, climate parameters and plant (crop) parameters. Generally, the required set of input parameters limits the use of LEACHM for research scale experimental plots; detailed site measurements cannot be obtained realistically for landscape or mapping unit scale applications. Recent access to digital soils and climate data bases (SCS, MAPS, Nielsen et al., 1990) coupled with pedotransfer functions (Rawls and Brackensick, 1989; Vereecken, et al., 1992) may be useful for screening soil mapping units for pesticide leaching potential. Howevery these efforts will require validation of transport model predictions within and between mapping units to determine the adequacy of transport predictions given different methods of obtaining input data. An important step in testing solute transport models under field conditions is a comparison of observed and predicted solute movement in controlled in situ experimental plots (e.g. research scale), where detailed measurements can be obtained for solute transport as a function of soil depth and time, and where input data can be obtained using independent estimates of soil, climate, chemical and plant input parameters. Solute transport studies were conducted at a field site near Manhattan, MT (Gallatin Co. Sec. 20, T IN , R3E) in a Brocko silt loam during the summers of 1991 and 1992. Studies were conducted to evaluate the fate and transport of several tracers and dicamba and to test the predictive capabilities of 3 LEACHM under field conditions. Two fluorobenzoates, pentafluorobenzoic acid (PFBA) and 2,6-difluorobenzoic acid (2,6-DFBA) (Bowman, 1984a; 1984b; Bowman and Gibbens, 1992), were used as nonreactive tracers, and in one study, PFBA was compared to Br" to evaluate differences in tracer performance under crop conditions where previous studies have shown that Br' is susceptible to plant uptake. Prior to our field work, a methodology study was conducted to develop analytical procedures for measuring fluorobenzoates simultaneously with Br using ion chromatography. The analytical procedure described in Chapter 2 is capable of analyzing aqueous solutions containing common soil anions (e.g. Cl", NO3", SO42"), orthotrifluoromethylbenzoic acid (o-TFMBA), 2,6-DFBA, PFBA, and Br' using an ion chromatograph equipped with an electrical conductivity (EC) detector (Pearson et a I.-, 1992). The objectives of these studies were to i) monitor in situ transport and fate of several conservative tracers (Br", PFBA, 2,6-DFBA) and a commonly used herbicide (dicamba) in a silt loam Montana soil over varying soil water regimes, under both cropped (barley - Hordeum vuigare L.) and fallow (bare soil) / management conditions and ii) evaluate the predictive capabilities of LEACHM (Wagenet and Hutson, 1989) given specific soil and climate input parameters for predicting the fate of these compounds under field conditions over varying soil water regimes. 4 CHAPTER TWO ANALYSIS OF FLUOROBENZOATE TRACERS BY ION CHROMATOGRAPHY Introduction Pentafluorobenzoic acid (PFBA), 2,6 difluorobenzoic acid (2,6 DFBA), and orthotrifluoromethylbenzoic acid (O-TFMBA) are three fluorobenzoates that have recently been used as tracers to monitor soil water and solute movement in soil systems. These fluorobenzoates are typically analyzed by high-performance liquid chromatography (HPLC) - UV detection and have reported retention times ranging between 9 and 12.5 mins. Ion chromatography (IC) analysis of these fluorobenzoates using conductivity detection is possible and faster than the \ currently used HPLC methodology. This study demonstrates the reliability of IC for fluorobenzoate tracer analysis. Mixed and single standard solutions (0 .10 to 25 mg L"1) of PFBA, 2,6 DFBA, O-TFMBA and LiBr were prepared and analyzed using IC-conductivity detection. High resolution among all four tracers was achieved by the methodology presented. However, Cl interfered with O- TFMBA analysis even at low concentrations (1.0 mg Cl L'1). Retention times of O-TFMBA, 2,6 DFBA, and PFBA by IC analysis ranged between 1.79 and 2.62 min with a linear detection response achieved over a 0.25 to 25 mg L"1 concentration range. These results indicate that IC analysis is an accurate and 5 expedient means of determining both single and multiple fluorobenzoate and Br concentrations in soil solutions and natural waters. Chemical tracers are commonly used to monitor soil water and solute movement in the vadose zone. By definition, (Davis et at., 1980) an ideal soil water tracer should be nontoxic, inexpensive, move with the water front, and be easy to detect in trace amounts. In addition, a conservative tracer should not alter the natural direction of water flow; it should be chemically stable and should not be sorbed or filtered by the solid medium through which the water moves (Davis et al., 1980). For most laboratory and field experiments, halides have been traditionally used as soil water tracers. As an alternative to halides. Bowman (1984a) introduced the use of fluorobenzoates as tracers for solute transport studies. Young and Boggs (1990) found that the fluorobenzoates, pentafluorobenzoic acid (PFBA), 2,6 difluorobenzoic acid (2,6 DFBA), and orthotrifluoromethylbenzoic acid (O-TFMBA) had mobility characteristics similar to Br. Of the three fluorobenzoates, PFBA and 2,6 DFBA have shown the most promise by demonstrating long-term resistance to chemical and biological degradation in the environment (Bowman and Gibbens, 1991). The current methodology used for fluorobenzoate analysis employs high-performance liquid chromatography (HPLC) coupled with variable wavelength ultraviolet (UV) photometric detection (Bowman, 1984b). Analysis times of PFBA, 2,6 DFBA, and O-TFMBA by this methodology are between 9 and 12.5 mins (Bowman, 1984b). C 6 The pKa's of PFBA, 2,6 DFBA and O-TFMBA range between 2.7 and 3.0 (Bowman and Gibbens, 1991). Consequently, these compounds are anionic at most soil pH's. The anionic nature of these fluorobenzoates allows for analysis by ion chromatography-conductivity detection using anion separator columns with low retention times. The need for expedient analysis of soil solutions containing these fluorobenzoates prompted the development of a method employing ion chromatography where retention and sample analysis times were less than current HPLC methodologies. Materials and Methods Standard solutions of PFBA, 2,6 DFBA, O-TFMBA (Aldrich Chemical Co., Milwaukee, Wl) and LiBr (Morton Thiokol Inc., Danvers, MA) were prepared from high purity compounds (> 98% ). Individual and mixed solutions of these fluorobenzoates and Br were prepared gravimetrically by carefully weighing the salts on an analytical balance and dissolving them in double deionized H2O. \ Solution concentrations included 0.10, 0.25, 0.50, 0.75, 1.0, 3.0, 5.0, 10, and 25 mg L"1. Analyses were performed on a Dionex 4000i ion chromatograph (Dionex Corp., Sunnyvale, CA) using a Dionex AS4A column with an eluting solution of 0.85 mM NaHCO3 and 6.90 mM Na2CO3. Electrical conductance detection levels used include: 3.//S (0.10, 0 .25, 0.50, 0.75 mg L"1 standard solutions); TO jjS (1.0 mg L"1); and 30 jjS (3.0, 5.0, 10 and 25 mg L"1). Chromatochart-PC (Interactive Microware, Inc. State College, PA) was used to integrate peak areas. All analyses were performed in triplicate. Retention times of the tracers were first determined using individual standards. Once this was accomplished, mixed standards of all four tracers were analyzed and linear plots of relative peak areas verses concentrations were developed. A 50 fj\ sample loop was used for solution analyses. Resolution comparisons were also made using a 100/vl sample loop. Results and Discussion Retention times of the mixed organic tracers and Br standard by IC analysis were found to range between 1.79 and 3.20 mins and are significantly less than the HPLC retention times presented by Bowman (1984b) (Table I) . Retention times of each tracer by IC analysis were slightly less when analyses were performed on individual standards (10 mg L"1) rather than mixed standards (TO mg L"1). This difference in retention times varied by no more than 0:59 mins (n =3). 7 8 Table I . Retention times of mixed organic tracers and bromide standard (25 mg L"1) using IC analysis and corresponding retention times using HPLC analysis. Tracer IC analysis retention time* HPLC analysis retention time min Bowman (1984b) min O-TFMBA 1.79 (0.00) 10.4 2.6 DFBA 2.05 (0.00) 12.5 PFBA 2.62 (0.01) . 9.0 Bromide 3.20 (0.01) 16.1 Values in parentheses indicate sample standard deviation of mean (n = 3). A mixed standard of Cl (1.0 mg L"1) and 25 mg L 1 of Br, NO3, and the three fluorobenzoates were analyzed by IC. Nitrate and Cl are commonly found in soil extracts and were added to the mixed tracer standard to demonstrate possible interferences. Ion chromatography analysis of the mixed standard indicated that high resolution among the four tracers was achieved by this methodology (Figure I) . Common soil anions that could be expected to interfere with tracer analysis included: Cl with O-TFMBA and NO3 with Br. Other anions commonly found in soil solutions such as PO4 and SO4 are eluted after the NO3 peak and will not interfere with tracer analysis. Nitrate interference with Br analysis is a common problem with HPLC analysis. The chromatograph illustrated in Figure I was generated using a Dionex AS4A column and demonstrates that NO3 did not interfere with Br analysis at the concentrations injected (25 mg L"1). Greater resolution between Br and NO3 9 can be achieved using a Dionex AS9 column and may be useful for analyzing Br solutions that contain high NO3 concentrations (data not presented). Although the AS9 column is useful for avoiding NO3 interferences, it has longer retention times for the inorganic anions than the AS4A column. Chloride interference with O-TFMBA is a problem even at Cl concentrations of 1.0 mg L'1. Resolution between Cl and O-TFMBA was not improved using the AS9 column (data not presented). ■' > i 10 L > . £ O O CN CL TIME (min Figure I . Mixed ion chromatograph of 3 fluorobenzoates, Br" and NO3, at 25.0 mg L 1 and Cl" at 1.0 mg L 1 using an output range detection of 30 /;S and an AS4A column. As indicated by the chromatograph peak areas (Figure I ), and the slopes of the peak areas versus concentrations (Figure 2), the sensitivity of the IC conductivity detector (with respect to peak area) was roughly 30 to 40 % greater for Br than for the organic tracers (by mass). Comparisons of all peak resolutions between the 50/vL and 10 0 //L sample loops indicated no difference when the total mass injected into the column was constant (data not presented). When the attenuation (conductance output range) was set to 3//S, peak areas of all fluorobenzoates and Br were reproducible by IC analysis at concentrations of 0,25 mg L"1, but not 0 .10 mg L'1. However, by lowering the attenuation to 0.3 jjS, peak areas of all tracers were reproducible at concentrations 0.025 mg L'1 (data not presented). Although a detection limit of 0.25 mg L 1 should suffice for tracer experiments, detection limits of 0.025 mg L"1 are attainable with careful analytical work and stable eluant baselines. Plots of concentrations versus peak areas for the three organic tracers and Br were developed to portray the linearity of the IC analysis (Figure 2). The coefficient of determination (R2) for relative peak area versus concentration (0.25 to 25 mg L"1) of the fluorobenzoates and Br were all greater than 0.99. The results presented (Figure 2) illustrate the linearity of the IC detector over three attenuations (3, 10, and 30 /vS). For greater accuracy, most analysts will want their standards to bracket their sample concentrations within one attenuation. 1 2 The influence of fluorobenzoates on seed germination and vegetative growth have not been previously reported. As a cautionary note to future users of fluorobenzoates on cropped experiments, we found that when PFBA (112 kg ha"1) was applied in conjunction with KBr (37 kg Br ha"1) in a field experiment, barley (Hordeum vulqare L.) growth was reduced by approximately 35% (observed at boot stage). We also observed in a laboratory experiment that barley seed germination was reduced in the presence of PFBA (112 kg ha" 1) and bromide (56 kg ha "1) (unpublished data) but not in the sole presence of either PFBA (112 Kg ha"1) or Br (56 kg ha 1). 2,6-OFBA PFBA G: 15 - 25 0 Concentration (mg L 1) Figure 2. Relative peak areas of O-TFMBA, 2,6-DFBA, PFBA and Br' versus concentrations using ion chromatography. Data points represent means of three replicates; all standard error bars fall within symbols. In summary, the fluorobenzoates PFBA, 2,6 DFBA and O-TFMBA plus Br were successfully analyzed by ion chromatography using conductivity detection. The IC methodology presented, provides an expedient and accurate means of determining fluorobenzoate concentrations. Retention and sample analysis times were found to be less than currently used HPLC methodologies. These results support the use of the IC method presented for both single and multiple fluorobenzoate analysis of soil solutions and natural waters. 13 14 CHAPTER THREE IN S ITU MEASUREMENTS AND LEACHM PREDICTIONS OF THE TRANSPORT OF PENTAFLUOROBENZOIC ACID AND BROMIDE IN A MONTANA SILT LOAM SOIL Introduction Solute transport models such as LEACHM (Wagenet and Hutson, 1989) / are becoming increasingly important tools for predicting, the fate of chemicals applied to soils. Deterministic transport models based on analytical or numerical solutions of the convective dispersion equation (CDE) yield unique predictions for a given set of site specific soil, chemical, vegetation and climatic properties. Such models are being used for a variety of purposes including education, screening pesticides for leaching potential, screening soil mapping units for leaching potential, and predicting the fate of chemicals at specific locations,; These models are also becoming popular with government agencies and environmental consulting firms for evaluating the fate of contaminants, and to support decisions regarding management practices which minimize the potential for leaching chemicals into shallow groundwater aquifers. Despite growing interest in the application of solute transport models such as LEACHM for predicting chemical movement in soils, there have been relatively few studies designed to verify model performance under field 15 conditions. Results of previous field studies (Fennel et al., 1990; Soulsby and Reynolds, 1992; Comfort et al., 1993; Jabro et al., 1993) have shown mixed results. Fennel et al. (1990) found that simulation results of Br" and aldicarb leaching for five models, including LEACHM, were similar. Based on a normalized objective function (NOF) data, these authors reported discrepancies of 30 to 45% between observed and predicted distribution of aldicarb and Br, and further stated that none of the models tested accurately predicted the distribution of Br and aldicarb in the soil profile. Jabro et al. (1993) compared ; movement of NO3-N to LEACHM predictions (N subroutine) and noted that, despite calibration of model parameters to site specific soil conditions, the model did not accurately predict NO3-N leaching at 1.2 m depth. Soulsby and Reynolds (1992) modeled soil water flux for an Al leaching study using LEACHM. They calibrated the model (i.e. optimized parameters such as Ks) using in situ tensiometer data, then compared model predictions against measured tensiometer data for the remainder of the year. Despite good convergence of simulated -and measured matric potentials throughout their calibration period, LEACHM predicted greater summer drying than actually occurred and required further optimization of Ks. In all of these studies, poor model performance was thought to result in part from the inability of LEACHM to simulate preferential flow. The criteria for evaluating the performance of solute transport models has been the subject of some debate. Fennel et al. (1990) suggested that the most 16 rigorous test of a model is a direct comparison of observed and predicted distribution of solute concentrations in the soil profile, and that pesticide simulation models should perform well enough to predict the center of solute mass and mass recoveries to within 50% of observed values. Smith et al. (1991) outlined Environmental Protection Agency (EPA) criteria for model acceptance. They noted that for screening applications with limited site- specific data and where model inputs have not been calibrated to the site, the model should be able to predict measured field data to within an order of magnitude. For site-specific applications where model calibration has been performed and on-site input parameters measured, the model should be able to match field data to within a factor of two. Smith et al. (1991) reported that the site specific criteria may be "too difficult to meet by even the best models using carefully measured site-specific parameters". Certainly, the predictive capability of deterministic models which do not accomodate preferential flow mechanisms may be limited to screening applications where such processes are important to water and chemical movement. A d hoc stochastic methods using deterministic models may be achieved by using confidence intervals of site-specific input parameters as representative of the inherent variability of soil properties within a given soil type. This approach was used by Comfort et al. (1993) in a field study of Br" transport under varying water regimes. Values representing upper and lower confidence intervals for independently measured soil retention coefficients (Campbell's 17 (1974) equation) were used as inputs to LEACHM, and reasonable agreement between observed and predicted soil water contents and Br" breakthrough curves (BTCs) was obtained. The study of Comfort et al. (1993) was conducted under fallow conditions on a spatially uniform Ioessal soil. The studies reported here (see also Chapter 4) for the 1991 and 1992 field seasons were conducted at the same site as those of Comfort et al. (1993), but were expanded to include transport of additional . nonreactive tracers (fluorobenzoates) and dicamba under both fallow and cropped conditions, under several water regimes. i Although Br' is widely considered as one of the best anionic nonreactive tracers (Davis et al., 1980), fluorobenzoates may be advantageous because their molecular weight and size is more similar to many organic solutes (Brusseau, 1993). Studies using fluorobenzoates (Bowman, 1984a; Bowman and Gibbens, 1992) have shown that pentafluorobenzoic acid (PFBA) and 2,6- difluorobenzoic acid (2,6-DFBA) are acceptable tracers for field studies. As these fluorobenzoates have pKa values less than 3.0, they are anionic at most soil pH's. Results from a multiyear aquifer tracer test (Young and Boggs, 1990) indicate that PFBA and 2,6-DFBA behave essentially identically to Br". Results from a laboratory study (Bowman, 1984a) indicate that PFBA, 2,6-DFBA and Br" behave similarly in repacked soil columns in the absence of plants. Although anion exclusion of nonreactive anionic tracers (Biggar and Nielsen, 1962; Thomas and Swoboda, 1970; Smith and Davis, 1974) may introduce 18 some discrepancy between water and solute movement (depending on soil type), a more important problem may be uptake of nonreactive tracers by plants. The uptake of Br" has been reported for potatoes (Solanum tuberosum L.), orchardgrass (Dacty/isglomerata L ) and Kentucky bluegrass (Poapratensis L.) during Br" transport experiments (Kung, 1990; Owen et al., 1985). If significant plant uptake of a tracer occurs, the mass of tracer in the soil volume is not constant, and if not accounted for, resulting changes in the shape of solute breakthrough curves (BTCs) may translate to erroneous interpretations of transport parameters. In addition, significant redistribution of tracer within the soil profile as a result of plant cycling can be especially problematic for long term tracer studies. Data from our laboratory has shown significant uptake of Br" by barley with subsequent release to the soil surface near plant senescence (unpublished data, in preparation). Hydraulic lift of soil water by roots from deep soil layers to surface soils having lower water potential has been documented in several studies (Cofak et al., 1987; Richards and Caldwell, 1987; Baker and van Baveb 1986) and may represent an additional mechanism for the redistribution of tracers which are susceptible to plant uptake. To our knowledge, there is no information currently available on whether fluorobenzoate tracers are taken up by plants to the same extent as Br". Nonreactive tracers not susceptible to plant uptake would be particularly advantageous for solute transport studies under cropped or natural ecosystems. 19 Specific objectives of this study were to (i) compare in situ transport of Br" and PFBA under cropped and fallow conditions over a range in soil water content and actual evapotranspiration, and (ii) evaluate the suitability of LEACHM using independent estimates of required input parameters for predicting solute movement over a wide range of environmental conditions. This study is the first of a two part series which is followed by an evaluation of observed and predicted transport of 2,6-difluorobenzoic acid (2 ,B-DFBA1) and the herbicide, dicamba. Materials and Methods A field experiment was conducted during the summer of 1991 to study in situ transport of nonreactive tracers under cropped [Hordeum vuigare L.) and fallow conditions at three different water regimes. Twenty-four PVC columns (0.20 m diameter, 1 .22 m depth) were installed in a Brocko silt loam (Borollic Calciofthid) at a field site near Manhattan, MT (Gallatin Co. Section 20, Township I 0N Range 3°E) (Table 2) using a hydraulic post pounder. The PVC .1 columns were fitted with steel or Al cutting heads to facilitate column insertion and to prevent soil compaction. Each column was equipped with three ceramic I suction cup Iysimeters (6 mm diam. X 80 mm length, air entry value -100 kP, Soil Moisture Equipment Corp., Santa Barbara, CA) at 0.36, 0 .66 and 0.96 m soil depths. Each Iysimeter was installed horizontally through the column wall by drilling a hole and removing a soil core of equal dimension to the sample cup. A slurry of silica flour was injected into the hole prior to inserting the 20 ceramic Iysimeters to ensure maximum soil-ceramic contact. The suction cup Iysimeters were attached to polypropylene tubing equipped with quick release couplers at the soil surface for solution sampling using dual port vacuum collection vials (20 ml). The columns were arranged in three rows of eight columns per row parallel to a line source irrigation system (Hanks et al., 1976), at distances of 1.5, 5.1 and 8.5 m, to establish decreasing soil water regimes (high, medium and low). Cropped and fallow treatments were imposed perpendicular to the line source irrigation system, resulting in four replications per water regime per cropped/fallow treatment. Adjacent fallow and crop treatments were separated by a 2 m fallow border and 2 m crop border. The crop treatments were seeded to barley [Hordeum vulgare L., cv. Klages) on May 24, 1991. On June 10, seventeen days after seeding, tracer solutions (100 ml) were applied uniformly to the surface of each column using an eyedropper. PFBA was applied at a rate of 112 kg ha"1 to all columns and KBr at a rate of 37 kg Br ha1 to only the crop columns. Bromide was not applied to the fallow columns because they had previously been used to study Br" transport during the summer of 1990 (Comfort et al., 1993) and may have contained some residual Br". Table 2. Profile characterization data of Brocko silt loam. Soil deoth Bulk densitv AEVt BCAMt Kr (I pH : IHoCM Organic matter Sand Silt Clay —m— 0.00-0.06 -M g m"3- 1.23 —kPa— -1.65 (.311)t mm d"1 4.64 (.227) 124(7.91) 8.1 9 Kg1 — — %)---- 0 .06-0 .14 1.29 -4.63 (.655) 3.87 (.192) 15.0 24 56 20 0.14-0.30 1.22 -2.70 (.270) 4.37 (.130) 377(94.0) ND§ 11.7 18 62 20 0.30-0.60 1.25 -9.72 (1.98) 1.65 (.157) 825(96.6) ND 4.0 22 67 11 0.60-0.80 0.80 T.12 1.31 1.32 -9.60 (1.69) -6.55 (.688) 1.63 (.134) 1.83 (.077) 585(229) ND 1.3 27 65 8 t AEV and BCAM determined by fitting the Campbell equation h = AEV(0v/0sat)"BCAM to soil water release data using nonlinear regression (0v determined at h values of 2.0, 5.0, 10.0, 20.0, 30.0, 50.0, 75.0 and 100 kPa). t Standard errors in parenthesis, § ND = Not determined. N> 22 Water was applied nine times over a 59-day period following chemical application (Table 3). Seven irrigations were applied with the line source system, followed by two manual applications resulting in a total of 51.6, 42 .0 and 31.1 cm of applied water (precipitation plus irrigation) for the high, medium and low water regimes, respectively. Soil water contents (0v, m3m'3) were measured throughout the experiment using a neutron moisture meter (CPN Corp., Martinez, CA), calibrated at this field site for converting neutron probe readings to volumetric water content (r2 = 0.95). Probe readings were taken adjacent to the in situ soil columns at 0.2, 0.4, 0.6, 0.8 and 1.0 m soil depths, with 2 replications for fallow and 3 replications for crop treatments for each water regime. Bare soilz / evaporation was measured in July over two independent wetting-drying cycles (5 replications per cycle) using 0.1 m diam by 0.2 m length minilysimeters (Lascano and Van Bavel, 1986). Sample Analysis Soil solution samples were collected 20 times during the 59-day period following chemical application, by vacuum extraction using a Cenco vacuum pump (Central Scientific Co., Chicago, IL) attached to the dual port collection vials (described previously). Minimal soil solution (approximately 3 ml) was collected from each Iysimeter during sampling, while providing adequate volume for ion chromatography analysis. To prevent contamination between sample dates, the first I ml collected was discarded and new vials were used to collect 23 the fresh samples. Soil solution samples were analyzed for PFBA and Br using a Dionex 40001 ion chromatograph equipped with an AS4A column (Dionex Corp., Sunnyvale, CA) (Pearson et al., 1992). Whole plant samples from the cropped columns were collected following the final soil solution sampling event, at Feekes growth stage 11.4 (Large, 1954). These were dried, weighed, ground and extracted using 25 mL 0.1 IVI NaNO3 per gram dry matter (Abdalla and Lear, 1975). The extracts were analyzed for Br" using an ion-specific electrode (model 9435BN, Orion Research Inc., Boston, MA), calibrated with Br' standards in a background matrix of extracting solution. Concentrations of PFBA and/or Br were plotted as functions of time for each Iysimeter depth (0.36, 0.66 and 0.96 m) to establish breakthrough curves (BTCs) for each chemical. Moment analysis (Skopp, 1984) of complete BTCs (primarily the 0.36 and 0.66 m depths) was used to estimate centers of mass (d), dispersion coefficients (cm2 d"1) and mean pore water velocities (mm d"1). Average 0v values from each Iysimeter depth were used to estimate mean soil water fluxes (mm d"1) and total water fluxes (mm) over the 59 d experiment. f The mass of PFBA and Br moving through the 0.36 and 0.66 m depths was then calculated based on the total water flux and individually measured PFBA and Br" concentrations. Percent recoveries for each chemical were calculated based on the fraction of applied chemical mass measured in the BTC at each Iysimeter depth. Table 3. 1991 Precipitation-irrigation and evaporation data used in LEACHM simulations. Precipitation-irrigation t F a llow Crop Evaporation D a t e t High • M e d iu m L o w High M e d iu m L o w W e e k W e e k ly pan — m m — — m m — - — m m — 1 0 Jun e (0) 2 2 . 5 2 1 . 4 1 3 .0 2 2 . 5 2 1 . 4 1 3 .0 I 4 5 . 5 1 4 June (4) . 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 2 4 8 . 3 1 9 June (9) 6 2 . 9 5 0 . 5 3 3 . 5 5 8 . 9 4 9 . 7 3 4 . 0 3 3 2 . 0 2 2 June (1 2 ) 1 .0 1 .0 1 .0. - 1 .0 1 .0 1 .0 4 4 6 . 0 - 2 3 June (1 3 ) 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 5 6 4 .5 2 5 Jun e (1 5 ) 8 1 . 9 6 1 . 5 4 1 . 6 7 1 . 8 5 8 . 8 4 0 . 0 6 5 3 .1 2 7 June (1 7 ) 2 . 0 2 .0 2 .0 2 .0 2 .0 2 . 0 7 6 4 . 8 2 8 Jun e (1 8 ) 2 . 0 2 .0 2 . 0 . 2 .0 2 . 0 2 .0 8 5 3 .3 3 0 Jun e (2 0 ) 4 . 0 4 . 0 4 . 0 4 . 0 4 . 0 4 . 0 9 4 7 . 5 2 Ju ly (2 2 ) 6 2 . 7 4 5 . 7 3 9 . 7 5 1 . 8 4 2 .1 3 5 .1 1 0 5 3 . 3 4 Ju ly (2 4 ) 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 0 Ju ly (3 0 ) 2 . 0 2 .0 2 . 0 2 .0 2 . 0 2 . 0 11 Ju ly . (3 1 ) 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 1 5 Ju ly (3 5 ) 6 5 . 9 5 0 . 8 4 2 . 3 5 6 . 9 4 5 . 7 4 3 . 6 1 9 Ju ly (3 9 ) 3 3 . 0 2 5 . 2 1 7 .6 2 8 . 8 1 9 .0 1 6 .8 2 0 Ju ly (4 0 ) 6 0 . 2 5 3 . 8 4 3 . 2 5 8 . 3 4 9 . 3 4 2 . 3 2 5 Ju ly (4 5 ) 3 . 0 . 3 . 0 3 . 0 3 .0 3 . 0 3 . 0 2 9 Ju ly (4 9 ) 6 4 . 5 5 2 .1 3 2 . 5 6 4 . 5 5 2 .1 3 2 . 5 . I A u g u s t (5 2 ) 1 .0 1 .0 1 .0 1 .0 1 .0 1 .0 5 A u g u s t (5 6 ) 6 3 . 5 5 1 .1 3 1 . 5 6 3 . 5 5 1 .1 3 1 . 5 ' T o ta l 5 3 6 .1 4 3 1 .1 3 1 3 . 9 4 9 6 . 0 4 0 8 . 2 3 0 7 . 8 5 0 8 . 3 t Values in parentheses indicate days after chemical application; dates are all 1991. t Rates of irrigation used for simulations were calculated based on actual application time. An average precipitation rate of 60 mm d"1 was used for all precipitation events. 25 LEACHM simulations Predicted BTCs for PFBA and Br' were generated using the LEACHP subroutine in LEACHM (version 2, Wagenet and Hutson, 1989) using I independent measurements or estimates of chemical, soil, climate and plant parameters (Tables 2, 3, 4). The source code array dimensions were modified to accommodate a greater number of depth nodes (56 nodes were used in our simulations). Soil physical parameters including bulk density, saturated hydraulic conductivity (KJ, and tp(6)qj-Qy relationships were determined as functions of soil depth using 4 replicate intact cores. Volumetric water contents (Sv) were determined at pressures of 2, 5, 10, 20, 30, 50, 75 and 100 kPa by placing intact soil cores in Tempe cells (SoiIMoisture Equip. Corp., Santa Barbara, CA). The resultant soil water release data were fit to Campbell's (1974) equation using nonlinear regression to obtain the air entry value (AEV) and exponent coefficient (BCAM) for model input (Table 2). The crop cover factor was estimated at 0.95 for the site, and the physiological maturity as defined by LEACHM was assumed to correspond to the booting stage, which occurred 58 d after seeding (41 d after chemical application). The uptake of nonreactive tracer (PFBA or Br") by plants was prohibited in the LEACHM simulations. 26 Table 4. Physical, chemical and crop input parameters used in LEACHM . simulations. Inout oarameter Unit Value Profile depth mm 1120 Segment thickness . mm 20 Boundary conditionf Molecular diffusion coefficientst § Do mm2 d'1 0.01 DIFFA 0.001 DIFFB 10.0 Dispersivity mm 3.2 PFBA and Br" properties'! Solubility mg L 1 9.8 X IO5 Vapor density mg L 1 0.0 K0=. L kg"1 0.0 Degradation constant d"1 0.0 Application rate mg m"2 PFBA 11200 Br 3700 Crop inputs , Crop cover 0.95 Date of ohvsioloaical maturitv# d 58 t Boundary condition 2 = unit gradient drainage. t From Bresler (1973). § Do: molecular diffusion coefficient in aqueous solution; DIFFA, DIFFB: constants a and b, respectively, in Dp = D0 a exp(/b0), where Dp is the effective diffusion coefficient. f K00: organic C distribution coefficient. # Expressed as days after seeding. Equivalent to 41 days after chemical application. ' * / 27 Results and Discussion Observed PFBA Transport Travel times required to detect initial solute fronts for PFBA (i) increased with soil depth under a given water regime, and (ii) increased with decreasing water application at a given soil depth (Figures 3, 4). Under fallow conditions, the leading edges of PFBA BTCs were detected within 8 ,1 1 and 18 d at 0 .36 m, within 17, 25 and 40 d at 0.66 m, and within 25, 40 and 51 d at 0 .96 m, for the high, medium and low water regimes, respectively (Figure 3). The travel times required to detect PFBA solute fronts at 0.36 m under high and medium water regimes were essentially identical under the cropped treatment (Figure 4). However, for the majority of BTCs at other depths and for the low water regime, slower and more disperse PFBA breakthrough occurred under cropped compared to fallow conditions. For example, under the Iowwater regime, PFBA ' r' solute fronts were detected at 0.36 m within 23 d for the crop treatment compared to 18 d for the fallow treatment. In addition, PFBA remained at 0 .36 m for considerably greater duration under cropped vs fallow conditions (Figures 3, 4). Travel times required to detect PFBA solute fronts were generally higher for the cropped treatments at 0.66 and 0 .96 m. For example, under the high water regime, PFBA solute fronts were detected at 0 .96 m within 46 d for the" crop treatment compared to 25 d for the fallow treatment. These observations suggest first that greater water use by the crop (ET) relative to fallow conditions (E) generally resulted in delayed PFBA transport and broader PFBA 28 BTCs. Secondly, effects of transpiration on solute transport were not significant at 0.36 m under high and medium water regimes because water application for these treatments far exceeded root water uptake early in the growing season during the time when PFBA was moving into the 0.36 m depth. Generally, the effects of root water uptake on solute transport were most significant at the lower depths (0.66 and 0.96 m) and under the low water regime. Although no PFBA was detected at 0 .66 m under cropped conditions under the low water regime, several samples (between 29 and 43 d after chemical application) could not be obtained due to the unavailability of soil water. Also, soil solution samples could not be obtained at 0.96 m where soil water contents dropped below 0.15 (Figure 6). Moment analysis (Skopp, 1984) was performed on those BTCs where solute mass returned to near baseline levels by 60 d after chemical application. Centers of mass (d) increased with increasing soil depth or with decreasing water application (Table 5). Under the high water regime, centers of mass (d) were significantly lower for fallow versus crop conditions at both the 0.36 and 0.66 m depths. Although this is consistent with the effects of crop water use on solute movement, the same trend was not ,observed under the medium water regime, where centers of mass were essentially identical for crop versus fallow conditions at both the 0.36 and 0.66 m depths. 400 300 200 100 High Wafer Regime (0.36 m) # Observed — LEACHM i t ? A lvv Medium Wate i r Regime (0.36 m) Low Water Regime (0.36 m) k T T — -r ‘ V ______ _ • / # X < 400- m u_ CL 300- | 200- _ l 0 )1 0 0 - High Water Regime (0.66 m) A A . Medium Water Regime (0.66 m) Low Water Regime (0.66 m) • • A 400 300 200 100 0< High Water Regime (0.96 m) ___________Afe- Medium Water Regime (0.96 m) Low Water Regime (0.96 m) )" " \ 0~~~20 30 40 50 60 70 C 10 20 30 *40 50 60 7 0 C 10 20 30 40 50 60 7C Days After Chemical Application Fig. 3. Observed and simulated PFBA concentrations at 0.36, 0.66 and 0.96 m depths under high, medium and low water regimes for fallow treatments. Vertical bars on symbols (observed data) indicate standard errors (n = 4), where absent, bars fall within symbols. CO O O IO 20 30 40 50 60 70 Days After Chemical Application Fig. 4. Observed and simulated PFBA and Br concentrations at 0.36, 0.66 and 0.96 m depths under high, medium and low water regimes for crop treatments. Vertical bars on symbols (observed data) indicate standard errors (n = 4), where absent, bars fall within symbols. 0.350 0.250 0.150 0.050 High Water Regime (1.0 m) 0 10 20 30 40 50 60 70 80 Medium Water Regime (0.40 m) Medium Water Regime (0.60 m) Medium Water Regime (1.0 m) 10 20 30 40 50 60 70 80 Days After Seeding Low Water Regime (0.40 m) t Low Wafer Regime (0.60 m) Low Water Regime (1.0 m) 0 10 20 30 40 50 60 70 80 Figure 5. Observed and simulated soil volumetric water contents at 0.40, 0.60 and 1.0 m depths under high, medium and low water regimes for fallow treatments, t indicates day of chemical application. Vertical bars on symbols (observed data) indicate standard errors (n = 2), where absent, bars fall within symbols. 0.350 High Water Regime (0.40 m) # Observed ----- LEACHM Medium Water Regime (0.40 m) 0.250 I_____— j 0.150 # • ^ 0.050 t f I £ 0.350 High Water Regime (0.60 m) Medium Water Regime (0.60 m) t^E . A K h % 0.250- I ^AiNvx -_______ ° 0.150 e # JD ^ 0.050 t t 0.350 0.250 0.150 0.050 High Water Regime (1.0 m) t Medium Water Regime (1.0 m) t O 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80 Days After Seeding Low Water Regime (0.40 m) Low Water Regime (0.60 m) t ' Low Water Regime (1.0 m) 10 20 30 40 50 60 70 80 CO N) Figure 6. Observed and simulated soil volumetric water contents at 0.40, 0.60 and 1.0 m depths under high, medium and low water regimes for crop treatments, t indicates day of chemical application. Vertical bars on symbols (observed data) indicate standard errors (n = 3), where absent, bars fall within symbols. Table 5. Moment analysis results based on selected^ observed and predicted _________ PFBA and Br breakthrough curves.________ • . ._________ ' _______ Soil Denth Cm) Water Reaime Treatment Center of : Observed PFBA Br mass(d) LEACHM PFBA D (cm2d"1 PFBA )§ Br Recovery(%) PFBA Br 0.36 High Fallow Crop 14.1 24.8 ND$ 16.6 21.5 22.5 4.74 4.93 ND . 2.92 116.2 86.8 ND 27.4 Medium Fallow Crop . 24.4 24.0 ND 20.6 27.0 ■ 26.0 2.17 2.70 ND 2.30 101.5 65.0 ND 22.6Low Fallow 29.6 ND 37.2 1.06 ND 67.0 ND0.66 High Fallow 26.8 ND 32.3 5.27 ND 91.9 ND Crop 37.6 29.0 36.2 3.12 2.08 120.5 14.3Medium Fallow 42.3 ND 43.8 1.67 ND 63.1 ND Crop. 39.3 34.3 45.5 ' 2.26 2.08 63.5 11.90.96 Hiah Fallow 38.7 ND 46.2 1.06 ND 61.3 ND f Values from truncated breakthrough curves do not represent true moments and are not presented. $ ND =■ Not determined - Br was not applied to fallow treatments. § D = dispersion coefficient. ww 34 Observed Br Transport and Plant Uptake Bromide was applied only to the cropped treatments (Figure 4) during the present study, because fallow columns had received a Br" application during the previous field season (Comfort et al., 1993). Travel times required to detect initial solute fronts for Br" were essentially identical to PFBA for all water regimes and for all depths where solute was detected. However, Br" BTCs at 0.36 and 0 .66 m consistently showed a loss of Br" relative to PFBA (Figure 4), beginning 25 to 30 d after chemical application (42 to 47 d after seeding). By this time, barley was at Feekes growth stage 10 (Large, 1954) and would likely have developed a functional root system to depths > 0.66 m (Borg and Grimes, 1986; Reid, 1985). Observed 0v's for the crop and fallow treatments were similar at 0 .4 and 0.6 m for any given water regime up to approximately 40 to 45 d after seeding (23 to 28 d after chemical application). After this S time, observed 0v's at 0 .4 and 0.6 m for the crop treatments declined significantly relative to the fallow treatment (Figures 5, 6), indicating root water uptake. The onset of significant crop water use at these depths corresponded to the time (± 2 d) when soil solution Br" began to decline relative to PFBA. By the time PFBA reached 0.96 m under the high and medium water regimes (45 d after chemical application and 62 d after seeding), Br" concentrations had declined significantly to near background levels, and Br" BTCs were essentially nonexistent. Interestingly, soil solution Br" reappeared several days after the original Br" BTCs reached baseline levels for several treatments (e.g. high water. 35 0.36 and 0.66 m and medium water, 0 .36 m). The reason for this observation is not completely understood, but may relate to Br cycling via plant roots. The differential response of PFBA and Br BTCs under cropped conditions observed in this study suggests that. PFBA may not be taken up by plants to the same degree as Br", and may thus be a more suitable tracer in the presence of plants. Further evidence of differential Br" uptake by barley relative to PFBA was obtained using moment analysis (Skopp, 1984) of the solute BTCs in the crop treatments (Table 5). As mentioned previously, PFBA and Br' BTCs were similar up to approximately 25 d after chemical application, after which Br concentrations declined significantly relative to PFBA (Figure 4). These changes in the shape of the BTC are reflected in the estimated centers of mass, dispersion coefficients, and the calculated percent recoveries of applied chemical detected under the solute BTCs obtained with moment analysis. Centers of mass (d) for Br' BTCs were consistently lower than PFBA BTCs due to the more rapid decline of Br" BTCs. In addition, dispersion coefficients at a given depth and water regime were lower for Br" compared to PFBA, again due to the earlier decline of Br" BTCs. Moreover, the percent recoveries of Br" calculated based on moment analysis ranged from 23 to 27 % at 0.36 m, and from 12 to 14 percent at 0.66 m. The decline in Br" recovery with increasing soil depth (and time) is consistent with the expected effects of increased cumulative plant uptake of Br with respect to time. More importantly, the Br" recoveries were significantly lower than calculated recoveries of 63 to 120 % 36 for PFBA BTCs under crop conditions. There appeared to be no significant difference in PFBA recoveries under crop versus fallow conditions. Although percent recoveries calculated from moment analysis are subject to some uncertainty due to the sensitivity of the calculation to variable solute concentrations (e.g. see error bars in Figure 4), it is clear that Br" recoveries were significantly lower than PFBA. This is consistent with the hypothesis that Br is taken up by plants to a much greater degree than PFBA. Concentrations of Br" in plant tissue were measured at harvest (Feekes growth stage 11.4, Large, 1954) and ranged from 1.52 to 1.22 g Br kg"1 for the high to low water regimes (Table 6). Concentrations of plant tissue Br" were converted to total Br' uptake per column using the measured dry matter produced per column. The total amount of applied Br" recovered in plant tissue at harvest ranged from 24% for the high to 18 % for the low water regime, indicating that significant Br' uptake occurred relative to the total mass of Br" applied to the soil surface. Other studies at this same field site have demonstrated significant plant uptake of Br under cropped {Hordeum vulgare L.) conditions. We have found that as much as 85 to 95 % of applied Br" (100 lbs acre"1) was taken up by barley by physiological maturity. However, between physiological maturity and harvest, Br" was lost from the above ground plant tissue and reappeared in the soil surface (0-6 cm). The measurements of Br" uptake obtained in this study were only obtained at harvest because we could not afford a sequential loss of plant tissue (the in situ columns have an area of 37 298.65 cm2). As a result, the percent recoveries of Br based on plant tissue analysis at harvest (Table 6) probably underestimated the amount of uptake which occurred earlier in the growing season when Br was moving through the soil profile. In fact, recoveries of soil solution Br" estimated using moment analysis suggest that approximately 80 to 90 % of the applied Br may have been taken up by barley between 42 and 62 d after seeding (Feekes growth stage 10, Large, 1954), whereas only 18 to 23 % of applied Br was accounted for in the plant tissue at harvest. These observations are essentially identical to our previous measurements of Br' uptake by barley with respect to plant growth stage obtained at the same site (unpublished data, in preparation). Finally, although the uptake of PFBA by barley was not measured directly, the peak concentrations of PFBA BTCs (Figures 3, 4) and total PFBA recoveries (Table 5) were similar under cropped vs fallow conditions. Consequently, our data suggests that PFBA was not subject to plant uptake to the same extent as Br' and may therefore be a more suitable nonreactive tracer for transport studies under cropped conditions. Table 6. Average Br" concentrations in whole plant (Hordeum vulgare L.) tissue samples and percent of applied Br" measured as plant uptake at harvest ________(Feekes growth stage = 11.4)$_______________ _̂_________• Water reoime a Br"(ko Dlant)"11 % Br" Recovery in Dlant masst high 1.52 (0.41) 23.35 (3.06) medium 1.36 (0.26) 20.89 (2.26) low 1.22 (0.25) 17.99 (2.04) t Standard errors in parenthesis (n =4). t Large (1954). A 38 LEACHM Predicted Soil Water Contents A comparison of observed and LEACHM predicted and soil volumetric water contents as a function of soil depth and time is critical to understanding differences between observed and LEACHM predicted water and solute fluxes. Under fallow conditions, predicted and observed 0v values were in reasonable agreement for all depths (0.4 to 1.0 m) throughout the majority of the growing season (Figure 5). However, under cropped conditions, LEACHM predicted 0v's were higher than observed in several cases (Figure 6). At 0 .4 and 0.6 m, LEACHM predicted 0v's were significantly higher than actual 0v's from approximately 45 to 55 d after seeding (Feekes growth stage 10, Large, 1954). At 1.0 m, LEACHM predicted 0v's were higher than observed during much of the growing season (Figure 6). Under cropped conditions, the root distribution function used in LEACHM plays an important role in determining predicted plant water uptake and subsequent water contents as a function of soil depth. The lack of agreement between predicted and observed 9v's suggests that this root growth and distribution function may not have adequately predicted root water uptake by the barley crop. As expected, LEACHM predicted cumulative evaporation plus transpiration was higher for crop versus fallow conditions (Table 7). However, predicted plant water uptake was heavily weighted within profile depths < 0.8 m. Roughly 57% of plant water uptake was predicted to occur from depths < 0 .4 m and 98% from profile depths < 0.8 m. Insufficient simulated water uptake may explain the higher LEACHM predicted 6v values relative to observed, especially at the 1.0 m depth. 39 Table 7. Observed and predicted water budget components from time of :_______ chemical application (t = 0) to final sampling event (t = 59)._____ Hiah Treatment Fallow Water regime Medium Low Hiah Croo Medium Low Observed Soil profile (t = 0) 248 245 -----mm----- 222 242 235 205 Cumulative precip./ irrigation 536 431 314 496 408 308 Cumulative drainaget 322 230 173 252 208 102 LEACHM Soil profile (t = 0 )t 237 235 222 232 227 209 Final profile (t = 5 9 ) . 318 306 279 311 292 247 Cumulative Evap. 184 178 93 . 54 54 50 Cumulative Trans. 0 0 0 183 184 185 Cumulative drainaoe 271 182 164 180 106 34 t Observed drainage calculations based on moment analysis from observed PFBA breakthrough curves. t LEACHM soil profiles at. time of chemical application (t = 0) differ from observed profiles at this same date due to the fact that LEACHM runs were initiated at time of planting (17 days prior to chemical application). LEACHM uses a root growth function based on corn in the midwest (Davidson et al., 1978), where a maximum rooting depth of I m was obtained by 86 d after seeding. Cereal crops grown in our region generally have a much shorter growing season and maximum rooting depths to > I m. Thus the mid and lower soil profile would likely experience plant water uptake sooner than predicted from the Davidson et al. (1978) model, and to a greater depth. Although this would potentially explain the lower observed soil water contents deep in the profile, one might also expect predicted water flow to be greater than observed due to higher conductivity with increased. soil wetness. However, the LEACHM predicted drainage component was far less than 40 calculated using moment analysis of the solute BTCs. Several additional factors other than the root distribution function may be important in explaining deviations between observed and predicted soil water contents and drainage components. First, soil retention coefficients (Campbell's equation) used to predict K(0) play a critical role in determining water flux and the availability of soil water for plant uptake within any given soil layer. Higher predicted 0v's suggest that the soil water retention function or our measured input coefficients overpredicted the water holding capacity of a portion of the soil profile. This would help explain why the drainage component predicted using LEACHM was less than that calculated using moment analysis. Second, it is important to note that the drainage component calculated using moment analysis is based on solute BTCs. Any preferential water or solute flow which may have occurred in the field is implicit in this estimate, whereas LEACHM does not currently incorporate preferential flow mechanisms. Deviation in predicted and observed Sv values at lower profile depths (0.4 to 1 .0 m) do not appear to be related to problems with predicted surface evaporation. Evaporation from the soil surface plays an important role in determining Sv values in the 0 to 0.2 m zone, and on subsequent water availability for transport deeper into the soil profile. Evaporation from bare soil was measured using minilysimeters (Lascano and van BaveL 1986) over two independent wetting-drying cycles, and showed excellent agreement with LEACHM predicted evaporation (Figure 7). Consequently, predicted evaporation 41 does not appear to be a source of error between observed and predicted soil water contents. Sensitivity analyses of LEACHM inputs indicated that changes in coefficients of Campbell's equation (BCAM and AEV), Ks, pan factor, and crop maturity date all influenced predicted 0v values. It would have been possible to obtain predicted 6v values which more closely matched observed values at lower profile depths by adjusting or fitting these parameters to the observed data. However, our objectives were to test predictions for transient water and solute flow under field conditions using a carefully measured set of input parameters determined in situ or using undisturbed soil cores. Moreover, any adjustments to the input parameters as required to improve agreement between observed and predicted 0v at lower soil depths resulted in poorer agreement at other depths (0.4 and 0.6 m). Medium Water Regime • Observed — LEACHM Date, July 1991 Figure 7. Observed and simulated daily surface evaporation from bare soil. 42 LEACHM Predicted BTCs Predicted PFBA BTCs (equivalent to predicted Br' BTCs) were close to observed BTCs in many cases (Figures 3 and 4). Under fallow conditions, predicted BTCs exhibited some shift to the right of observed BTCs for the high and low water regime. Centers of mass for observed and predicted PFBA BTCs (Table 5) were 14 and 22 d at 0.36 m, 27 and 32 d at 0.66 m, and 39 and 46 d at 0 .96 m. Although greater preferential flow might be expected at higher Ov values and under higher pore water velocity (Beven and Germann, 1982) no consistent trends were noted in our data for early solute breakthrough (relative to predicted) versus water application levels. Even the low water regime received approximately 30 cm of total water during the growing season, consequently, preferential flow may have been feasible in all of the water regimes studied. Under cropped conditions, predicted PFBA BTCs (Figure 4) and centers of mass (Table 5) were generally very closie to observed values. No significant evidence of preferential flow was observed under crop conditions, perhaps with the exception of the 0.96 m depth under the medium water regime (Figure 4). In fact, LEACHM predicted BTCs were shifted to the left of observed BTCs in several cases. For example, in the low water regime, the observed PFBA BTC at 0.36 m shows signs of delay due to crop water use, whereas the LEACHM curve does not. At 0 .66 m the LEACHM predicted BTC ■ appears to be far ahead of any measurable PFBA. Finally, it should be noted that predicted PFBA BTCs were generated assuming no solute uptake by plants. 43 Visual inspection of observed and predicted BTCs under cropped conditions (Figure 4) shows that this appears to be a much better assumption for PFBA than for Br. Conclusions As expected, in situ measurements of the transport of PFBA (and Br" under cropped conditions) demonstrated increasing solute travel times with increasing soil depth and decreasing water application. The presence of barley in the cropped treatments caused measurable delays in the transport of PFBA, especially under the low water regime where the total amount of applied water (31 cm) did not greatly exceed plant consumptive demand. Independent estimates or measured values of soil, climate, and plant parameters were used as input data to the transport simulation model, LEACHM. Comparisons between, measured and LEACHM predicted 0v's at 0.4m and bare soil surface evaporation were generally in good agreement. Predicted 0v's-at lower profile depths (0.6 to I .Om), particularly under cropped conditions, were often higher than measured values. This appears to be a.result of inadequate assumptions in LEACHM concerning the root growth function, which influences the amount of predicted plant water uptake from a given soil layer. Predicted breakthrough curves (BTCs) and centers of mass (d) of the nonreactive tracer, PFBA, were generally in good agreement with measured values. Under cropped conditions, the transport of Br" and PFBA was similar up to about 40d after seeding. After this time, Br" BTCs showed a marked 44 decension relative to PFBA BTCs, as a result of plant uptake of Br". Consequently, our data suggest that PFBA is a more suitable nonreactive, conservative tracer than Br". Validation of transport models under field conditions is necessary for : \ verifying model predictive capabilities under a wide range of environmental conditions. LEACHM is a research oriented simulation model requiring extensive inputs. Our evaluation of LEACHM supports the use of this model based on the criteria presented by Pennel and others (1990), i.e., predicted centers of solute mass are within 50% of observed values. Furthermore, simulation results nearly met the EPA criteria for site specific, calibrated models where predicted profile concentrations are required to be within a factor of two, despite the fact that we did not attempt to calibrate LEACHM to our specific soil hydraulic conditions. v 45 CHAPTER FOUR IN SITU MEASUREMENTS AND LEACHM PREDICTIONS OF THE TRANSPORT AND FATE OF 2,6-DIFLUOROBENZOIC ACID AND DICAMBA IN A SILT LOAM MONTANA SOIL Introduction Pesticide leaching and subsequent contamination of groundwaters from land under agricultural production is becoming a significant national concern. Recent monitoring programs have detected over 70 pesticides in groundwaters of 38 states (Ritter, 1990; Parsons and Witt, 1988). In one of thesb surveys, 17 pesticides were detected at concentrations above health advisory limits. DeLuca et al. (1989)' and Clark (1990), conducted well water monitoring surveys in agricultural areas in Montana and documented that several pesticides (including aldicarb, atrazine, 2,4-D, dicamba, MCPA, picloram and simazine) have migrated into shallow groundwaters, presumably through normal agricultural management practices. Dicamba (3,6-dichloro-2-methoxybenzoic acid) is commonly used in Montana to control broadleaf weeds in small grain production systems and has been found in groundwater samples regionally (DeLuca et al., 1989). Dicamba has a pka of 1.95, is anionic at ambient soil pH's (Weber, 1977) and is highly soluble in water (6.5 X 1.0® mg L"1) (Pesticide Manual - 9th ed., 1991). 46 Consequently, dicamba has a low affinity for soil colloids (Burnside and Lavy, 1966; Grover and Smith, 1974; Jury et al., 1987) and is considered highly mobile in most soils (Friesen, 1965; Grover, 1977; Scifres and Allen, 1973). Soil environmental conditions play a significant role in the fate and mobility of dicamba applied to soils. Degradation rates of dicamba in surface soils, are generally fairly rapid with half-lives ranging from 13.5 to 45 days (Comfort et al., 1992; Krueger et al., 1991; Smith, 1974). The primary metabolite of dicamba degradation is 3,6-dichlorosalicylic acid (DCSA), which is more persistent and not nearly as mobile as the parent compound (Smith, 1973a; Smith, 1974). Consequently, the potential for dicamba to leach out of the rooting zone increases dramatically if soil conditions limit its degradation rate. Conditions which limit the degradation rate of dicamba in soils include low soil organic C content (Smith, 1973b), low soil temperature (Smith and Cullimore, 1974; Comfort et al., 1992), soil pH > 7.5 (Corbin and Upchurch, 1966) and high water application rate (Comfort et al., 1992). These conditions may result in the transport of dicamba out of the root zone and into shallow groundwaters. In a laboratory column and incubation study. Comfort et al. (1992) showed that transport and subsequent loss of dicamba out of soil columns could be substantially reduced by delaying water application, thus providing sufficient time for the formation of DCSA. The current study was designed in part to broaden our understanding of the fate of dicamba in field soils under a range of soil water and evapotranspiration conditions (i.e., crop vs. fallow). 47 In response to regional concerns about the potential mobility of broadleaf herbicides under irrigated conditions we have been studying the fate and mobility of 2,4-D and dicamba under different soil water regimes, under both laboratory (Comfort et al., 1992) and field conditions. In addition, we are interested in potential applications and limitations of solute transport modeling for indexing soil mapping units according to their susceptibility for pesticide leaching (Wilson et al., 1993). As part of an overall process to evaluate the suitability of transport models for predicting solute transport under field conditions at landscape scales, models should be tested against observed data using input data sets generated from different scales of observation. This study was conducted at a research plot scale using a fairly intensive set of independently measured model input parameters. The specific objectives of this study were to (i) monitor the transport and fate of dicamba and a nonsorbing, nondegrading tracer, 2 ,6-difIuorobenzoic acid (2,6-DFBA) under fallow (bare soil) and cropped (barley - Hordeum vulgare L.) conditions and varying soil water regimes, and (ii) evaluate the suitability of LEACHM (Wagenet and Hutson, 1989) for predicting the fate of these compounds under field conditions, given specific soil and climate input parameters. This is the second of two manuscripts (see Chapter 3) which evaluate the transport of nonreactive tracers and dicamba bver a range of applied water and evapotranspiration conditions. 48 Materials and Methods Field study A field experiment was conducted to study in situ transport of 14C ring- labeled dicamba (Sigma Chemical Company, St. Louis, MO) and 2,6- difluorobenzoic acid (2,6-DFBA) (Aldrich Chemical Co., Milwaukee, Wl) during the summer of 1992. Eighteen of the 24 PVC columns (0.20 m diameter, 1.22 m depth) described in the companion paper (Chapter 3) were used in the current study on a Brocko silt loam (Borillic Calciorthid) near Manhattan, MT (Gallatin Co. Sec 20, T IN , R3E). Three, rows of six columns per row were positioned parallel to a line source irrigation system (Hanks et al., 1976) at distances of 1.5 and 5.1 and 8.5 m to establish decreasing soil water regimes (high, medium and low). Two treatments, crop (barley, Hordeum vuigare L.) and fallow, were imposed at each soil water regime, with three replications per treatment. The fallow treatment columns were installed in 1990 and had been I previously used in Br and pentafluorobenzoic acid (PFBA) transport experiments. The crop treatment columns were originally installed in 1991 and had been previously used in PFBA and Br transport experiments under cropped conditions. The'crop treatments were seeded to barley (Hordeum vuigare L. - cy. Klages) on May I , 1992. Solutions (100 ml) containing dicamba (14C-ring labelled) and 2,6-DFBA (nonreactive tracer) were uniformly applied drop wise to the surface of each soil column on June 18, 1992 (48 d after crop treatments 49 were seeded). Dicamba was applied at a normal field application rate of 0.26 kg ha'1 (2.04 X IO 4 mCi 14C kg'1) and 2,6-DFBA at 112 kg ha'1 (Table 8). Dicamba was applied only to the high and medium water regime columns. Water was applied eight times to the experimental site over a period of 67 days following chemical application (Table 9). Six irrigations were made using the line source irrigation system plus two final applications by hand, resulting in a total water application (irrigation plus precipitation) of 40.6, 34.4, and 27.8 cm to the high, medium, and low water regimes, respectively. Soil solution samples were vacuum extracted 27 times during the experiment to monitor the transport of 2,6-DFBA and 14Odicamba. Minimum soil water was collected during each sampling event (approximately 3 to 5 ml per lysimeter) to minimize soil solution disturbance in the vicinity of the lysimeters. Volumetric soil water contents (0V, m3m"3) were monitored at each sampling date by neutron attenuation using a Campbell hydroprobe (CPN Corp., Martinez, CA). A field calibration was used to convert neutron meter readings to volumetric water content (r2 = 0.94). Probe readings were taken adjacent to the in situ soil columns at 0.2, 0.4, 0.6, 0 .8 and 1.0 m depth increments with two replications for the fallow treatment and three replications for the cropped treatment at each water regime. Bare soil evaporation was measured using 10 cm dia. by 20 cm length minilysimeters (Lascano and van Bavel, 1986) in late July, 1992 over two independent wetting-drying cycles. 50 Table 8. Physical, chemical and crop input parameters used in LEACHM simulations. Inout oarameter Unit Value Profile depth mm 1120 Segment thickness mm 20 Boundary conditiont Molecular diffusion coefficientst § Do mm2 d'1 0.01 \ DIFFA 0.001 DIFFB 10.0 Dispersivity mm 3.2 Chemical properties Solubility mg L"1 Dicamba 6.5 X IO 3 DCSA 6.5 X IO 3 2,6-DFBA 9.8 x I O5 Vapor density mg L 1 Dicamba > 4.1 X 10-4 DCSA 4.1 X IO"4 2,6-DFBA 0.0 Kocf L kg’1 Dicamba 0.0 DCSA 5.0 X IO 2 2,6-DFBA 0.0 Degradation rate constant d'1 Dicamba 5.1 X IO 2 DCSA 1.7 X IO 2 2,6-DFBA 0.0 Application rate mg m"2 Dicamba 23.15 2,6-DFBA 11200 Crop inputs Crop cover 0.90 Date of ohvsiolodical maturitv# d 70 t Boundary condition = 2 unit gradient drainage. t From Bresler (1973). § Do: molecular diffusion coefficient in aqueous solution; DIFFA, DIFFB: constants a and b, respectively, in Dp = D0 a exp(b9), where Dp is the effective diffusion coefficient, f K00: organic C partition coefficient. # Expressed as days after seeding. Equivalent to 22 days after chemical application. Table 9. Precipitation-irrigation and evaporation data used in LEACHM simulations. Precipitation-irricrationl Fallow________;_____________ Crop Evaporation . Datef Hiah Medium Low Hiah Medium Low Week Weeklv pans ---mm—-— -- mm--- ' ---mm--- 18 June (0) 20.5 17.7 ' 9.7 20.7 17.0 9.9 I 48.2 23 June (5) 44.3 33.4 22.1 41.9 33.5 23.3 2 35.2 . 24 June (6) 1.0 1.0 1.0 . 1.0 1.0 1.0 3 38.2 25 June (V) 2.0 2.0 2.0 2.0 2.0 2.0 4 35.8 "I July . (13) 1.0 1.0 1.0 1.0 1.0 1.0 5 38.6 2 July (14) 14.0 14.0 14.0 . 14.0 14.0 14.0 6 53.2 3 July (15) 1.0 , I -; 0 1.0 1.0 1.0 1.0 7 - 55.4 4 July (16) 2.0 2.0 2.0 2.0 2.0 2.0 8 52.8 5 July (17) 1.0 1.0 . 1.0 1.0 1.0 1.0 9 53.6 6 July (18) 12.0 12.0 12.0 12.0 12.0 12.0 10 38.3 I July (19) 33.9 26.5 17.9 32.6 24.4 17.1 13 July (25) 3.0 3.0 3.0 3.0 3.0 3.0 14 July (26) 42.0 35.8 25.0 35.7 33.6 24.0 21 July (33) 45.5 37.8 26.9 41.3 35.3 28.6 23 July (35) 4.0 4.0 4.0 4.0 4.0 4.0 28 July (40) 38.8 35.2 27.3 38.0 32.9 23.0 11 Aug. (54) 63.5 50.7 38.1 63.5 50.7 38.1 19 Aug. (60) 50.7 38.1 38.1 50.7 38.1 38.1 2I Aug. (62) 23.0 23.0 23.0 23.0 23.0 23.0 22 Aug. (65) 1.0. 1.0 1.0 1.0 1.0 1.0 23 Aug. (66) ■ 6.0 6.0 6.0 6.0 6.0 6.0 24 Aug. (67) 3.0 3.0 3.0 3.0 .3.0 3.0 Total 413.2 349.4 279.1 398.5 339.5 276.2 449.3 f Values in parentheses indicate days- after chemical application; dates are all 1992. $ Rates of irrigation used for simulations were calculated based on actual application time. An average precipitation rate of 60 mm d"1 was used for all precipitation events. § Daily pan evaporation was collected on site from a Class A National Weather Service .evaporation pan. 52 Sample Analysis All soil solution samples were analyzed for total 14C using a Packard 2200 CA liquid scintillation analyzer. Selected samples were analyzed for dicamba and metabolites using an HPLC equipped with a Lichrosorb RP-18 column (EM Science, Gibbstown, NJ) and Beckman model 171 radioisotope detector (Beckman Instruments Inc., Fullerton, CA). In all cases, soluble 14C was identified as ,dicamba with no detectable levels of the principle metabolite, DCSA. All Iysimeter samples were analyzed for 2,6-DFBA using ion chromatography - electrical conductivity detection (Dionex 4000i) with a Dionex AS4a column (Dionex Corp., Sunnyvale, CA) following the procedures outlined in Pearson et al. (1992). On August 27, 1992 all soil columns were removed and later sectioned in 0.1 m depth increments. The soil was dried at 37°C , then ground to pass through a 2 mm sieve. A 1.0 g subsample from each 0.10 m depth increment was oxidized with a model 0X 300 Biological Oxidizer (Model 0X 300, R.J. Harvey Instrument Corp., Hillsdale, NJ) and analyzed for total 14C using liquid scintillation analysis. In addition, selected samples based on the distribution of total 14C with respect to soil depth were solvent extracted (Smith and Muir, 1980) to determine the amount of 14C present as dicamba or secondary metabolites. The extracting solution contained 70% acetonitrile, 27% H2O, and 3% glacial acetic acid. Fifty grams of soil were combined with 1.00 mis of extracting solution and were shaken for 12 hrs. After agitation, the suspension 53 was filtered and 50 mis of the filtrate were combined with 100 mis of IM HCI and successive additions of 26 and 13 mis of methylene chloride (CH2CI2). A separatory funnel was used to collect the CH2CI2 fraction which was then evaporated to I ml. These solutions were analyzed using HPLC-radioisotope detection (described previously). Percent recoveries of spiked soil samples for dicamba and DCSA using this methodology were 86% and 47% , respectively. Concentrations of 14C-dicamba and 2,6-DFBA were plotted as functions of time for each Iysimeter depth (0.36, 0.66 and 0.96 m) to establish breakthrough curves (BTCs) for each chemical. Moment analysis (S.kopp, 1984) of complete BTCs (primarily the 0.36 and 0.66 m depths) was used to estimate centers of mass (d), dispersion coefficients (cm2d'1) and average pore water velocities (mm d"1). Average Qy values at each Iysimeter depth were used to calculate average water fluxes (mm d"1) and total water fluxes (mm) over the 67 day experiment. The mass of dicamba and 2,6-DFBA moving through the 0.36 and 0.66 m depths was then calculated based on the average total water flux and individually measured dicamba or 2,6-DFBA concentrations. Percent recoveries for each chemical were calculated based on the fraction of applied chemical mass measured in the BTC at each Iysimeter depth. LEACHM Simulations Predicted BTCs for 2,6-DFBA and dicamba were generated using the i LEACHP subroutine in LEACHM (version 2, Wagenet and Hutson, 1989) with independent measurements or estimates of chemical, soil, climate and. plant 54 parameters (Tables 8, 9). Soil physical and hydraulic input parameters were described in detail in a companion paper (e.g. see Table 2, Chapter 3). Batch adsorption studies were used to obtain independent estimates of Koc for dicamba and its principle metabolite, DCSA. Duplicate soil suspensions (10 grams soil from 0 to 15 cm: 10 mis H2O) were spiked with 14C ring labeled dicamba or DCSA at concentrations ranging from 0 .24 to 2.25 mg L 1 (4.56 X IO 7 to 4.56 X 10-5mCi 14C ml"1) and 0.01 to 0.70 mg L 1 (4.81 X 10"7 to 4.81 X 10"5mCi 14C ml"1) for dicamba and DCSA, respectively. These concentration ranges bracketed the levels of dicamba and DCSA measured in Iysimeter soil solutions during the field experiment. The soil suspensions were shaken for 48 hrs, then centrifuged at 2000 g for 15 min. A I ml aliquot of the supernatant solution was used to determine 14C activity by liquid scintillation analysis (defined previously), and the amount of DCSA or dicamba sorbed was determined by difference between total added and that remaining in the supernatant. The arnount of dicamba sorbed was not statistically different than 0 (i.e., Koc= O') over the range of dicamba levels investigated, while the sorption coefficient (Koc) for DCSA was 504 L kg"1. First-order degradation rate constants for dicamba and DCSA for the soil surface horizons were estimated from previous batch studies in our laboratory (Comfort et al., 1992) and from published values (Smith, 1974; Krueger et al., 1991). Values of 0.051 d"1 (t1/2= 13.5 d) for dicamba and 0 .017 (t1/2= 40 d) for DCSA were used for the soil surface horizon (0 to 30 cm). Degradation 55 rates of herbicides often decrease with soil depth as a result of reduced microbial numbers or microbial activity (Moorman and Harper, 1989; Pothuluri et al., 1990; Veeh and lnskeep, 1990). Rate constants for soil depths greater than 30 cm were calculated using an exponential decay function as outlined in Jury et al. (1987): t1/2(depth) = t1,/2(surface)el'(z'L) where y is a depth constant (1 .5 m'1), Z is depth to the ith layer and L is the depth of the surface layer. Although these are estimated values we have found that this function provides a reasonable estimate of the variation in 2,4-D degradation as a function of soil depth (Veeh and lnskeep, 1990) and the behavior of organic matter (which often relates to microbial activity) as a function of soil depth (Wilson et al., 1993). Results and Discussion Observed Dicamba and 2,6-DFBA Breakthrough Curves As expected, travel times required to detect solute fronts for dicamba and 2,6-DFBA increased with increasing soil depth at all water regimes, under crop and fallow conditions (Figures 8 to 11). At a given sample depth, the travel time required to detect solute fronts generally increased with decreasing, water application; this observation was most consistent for solute BTCs on the fallow treatments (Figures 8, 10). For example, dicamba fronts (Figure 8) were detected within 5 and 15 d at 0 .36 m and within 26 and 39 d at 0.66 m for the high and medium water regimes, respectively (14C-IabeIed dicamba was not i 56 applied to the low water regime treatment). The leading edges of 2,6-DFBA BTCs were detected within 5, 13 and 22 d at 0.36 m and within 26, 36 and 57 d at 0.66 m for the high, medium and low water regimes, respectively (Figure 10). Under cropped conditions, very little difference in travel times was observed for solute BTCs under high and medium water regimes (Figures 9, 11), and travel times were significantly shorter at a given depth than under fallow conditions. For example, 2,6-DFBA and dicamba fronts were observed within 5 d at 0 .36 m and at 22 d at 0.66 m for both the high and medium water regimes (Figures 9, 11). Even the 2,6-DFBA front for the low water regime at 0.36 m had a shorter travel time under crop (16 d) vs fallow (22 d) conditions. Results from moment analysis (Skop'p, 1984) are presented (Table 10) only for those solute BTCs where solute concentrations returned to near baseline levels by 67 d after chemical application (last day of Iysimeter sampling). Estimated centers of mass (d) for dicamba and 2,6-DFBA BTCs ranged from 21 to 29 d at 0.36 m and 41 to 47 d at 0.66 m. Average centers of mass appear to be slightly greater for 2,6-DFBA than dicamba for all depths, water regimes and crop vs. fallow treatments. Measured Koc values for dicamba for the surface (0 to 15 cm) layer were essentially 0 for this soil. Although it is assumed that Koc values for 2,6-DFBA are also 0 based on other studies (Bowman, 1984a, 1984b; Bowman and Gibbens, 1992; Young and Boggs, 1990), it is possible that given such low sorption for dicamba, 2,6-DFBA may 57 be sorbed slightly more than dicamba. Dicamba -and 2,6-DFBA BTCs consistently showed increasing centers of mass with decreasing water regimes for the fallow treatment (Table 10). However, for the cropped treatment, centers of mass are essentially identical at high and medium water regimes (i.e. 21 and 20.5 d for dicamba at 0.36 m, and 41 .4 and 43 d for dicamba at 0.66 m under high and medium water regimes, respectively). Finally, 2,6-DFBA BTCs consistently showed lower centers of mass at a given depth and a given water regime under crop vs fallow conditions (Table 10). Originally, we expected that the transport of dicamba and 2,6-DFBA would be delayed considerably under cropped conditions, similar to PFBA transport observed in our previous study (Pearson et al., this issue). However, despite higher evapotranspiration (ET) in crop treatments, dicamba and 2,6- DFBA transport was as fast, if not faster under crop conditions. Two factors may have been responsible for this observation. First, evaporation from the soil surface (0 to 15 cm) would likely be greater under fallow than crop conditions, allowing a greater fraction of applied water to move more deeply into the soil profile under crop conditions (discussed further in section on LEACHM predictions). Secondly, there may have been differences in temporally variable soil hydraulic properties related to differences in macroporosity between the crop and fallow columns. The Brocko silt loam was formed from Ioessal parent material and the spatial variability of soil properties at this field site is rather low. Soil water retention relationships for four replicate intact soil cores taken 58 near the in situ columns show fairly low variability; standard errors of the exponent in the Campbell equation were typically less than IO percent (Table 2, Chapter 3). Consequently, it is highly unlikely that spatial variability would have coincidentally biased the in situ columns used for either the crop or fallow treatments. More importantly, the columns used for the fallow, and crop treatments during this field season had slightly different management histories. The fallow columns were installed in 1990 and had been in fallow consecutively since then. Thecrop columns were installed in 1991 and had one year of barley prior to this study. Results from the previous study with PFBA transport (Chapter 3) showed no consistent evidence of any preferential water flow. However, decaying or active root channels might increase the potential for preferential solute movement in columns under crop conditions. Results from studies conducted by Meek et aL (1992) and Zins et al., (1991) have shown that the presence of alfalfa roots increase infiltration rates through soils. Beven and German (1982) suggested that preferential flow pathways are associated with live and decayed roots. Gish and Jury (1983) reported dispersivities of Cl transport through a sandy loam soil as influenced by dead wheat root systems. The dispersivities they observed were characteristic of dispersivities obtained in a "porous medium with significant flow occurring in large pores". In fact, moment analysis of our data showed that dispersion coefficients at 0 .36 m for both 2,6-DFBA and dicamba were roughly 3 times higher in cropped than fallow columns (Table 10) for both the high and medium 59 water regimes. Given that the cropped and fallow columns were subjected to nearly identical water applications, one explanation for the observed increases in solute dispersion in the crop treatments is a greater heterogeneity of pore water velocities resulting from root channels. Moment analysis was also used to estimate the percent of applied solute measured in individual solute BTCs (Table 10). Percent recoveries ranged from 91 to 150 % for 2,6-DFBA and from 42 to 117 % for dicamba over all treatments where analysis of complete BTCs was possible. Values greater than 1.00% are obviously not possible; given the variability of solute concentrations in replicate columns as functions of time (see error bars in Figures 8 to 11), estimated recoveries should not be considered absolute. However, the percent recoveries for 2,6-DFBA BTCs do suggest that the Iysimeters adequately sampled the solute breakthrough at 0.36 and 0.66 m, and that significant bypass flow around the Iysimeters did not occur. In addition, the fact that recoveries for dicamba were roughly 20 to 40% lower (average of 33%) than recoveries for 2,6-DFBA indicates that degradation of dicamba was occurring during transport. 0.8 0 .3 0 0 0.200 0.100 0.000 High Water Regime (0.36 m) • Observed — LEACHM T•<» • •i I1 jj 0 .3 0 0 0 .2 0 0 0.100 0.000 High Water Regime (0.96 m) 10 20 3 0 40 50 60 70 0 10 20 30 40 50 60 70 Days A f te r Chem ica l App lica t ion O)O Fig. 8. Observed and simulated dicamba concentrations at 0.36, 0.66 and 0.96 m depths under high and medium water regimes for fallow treatments. Vertical bars on symbols (observed data) indicate standard errors (n = 3), where absent, bars fall within symbols. • Observed — LEACHM High Woler Regime (0.36 m) Medium Water Regime (0.36 m) 0 .3 0 0 0.200 0.100 * • *.** *,** •+ i * * High Water Regime (0.66 m) Medium Water Regime (0.66 m) 0 .3 0 0 0.200 0.100 High Water Regime (0.96 m ) Medium Water Regime (0.96 m) 0 .3 0 0 0.200 0.100 Days After Chemical Application Fig. 9. Observed and simulated dicamba concentrations at 0.36, 0.66 and 0.96 m depths under high and medium water regimes for crop treatments. Vertical bars on symbols (observed data) indicate standard errors (n = 3), where absent, bars fall within symbols. High Water Regime (0.36 m) • Observed -----LEACHM High Water Regime (0.66 m) I 300 High Water Regime (0 .96 m ) O 10 20 30 40 50 60 70 Medium Water Regime (0.36 m) Medium Wafer Regime (0.66 m) Medium Water Regime (0.96 m) Days After Chemical Application Low Water Regime (0.36 m) Low Water Regime (0.66 m) Low Water Regime (0.96 m) ONJ Fig. 10. Observed and simulated 2,6-DFBA conc